WO2017033292A1 - System and method for stabilizing power control - Google Patents

Abstract

Non-server/client-type independent-distribution parallel processing control systems significantly reduce the communication load in the control system and make rapid response possible even in large-scale systems. Despite this, there has been little research and discussion with regard to the impact on the stability of the overall system from delays in the dynamic characteristics or output information of each unit and transmission delays in the broadcasting system in such a method, and with regard to an active stabilization method therefor. Establishing stability can be difficult even in simple broadcast-type systems in which the priority is not different for each unit. An independent-distribution system stabilization method resulting from the conditions for stability evaluation and stability limits and the necessary conditions for the "characteristic priority", i.e. the direct current gain, provided in the present invention, as well as from equipping each unit with a dynamic compensator, is effectively provided for the first time by the present invention.

Description

System and method for stabilizing power control

The present invention is a control system constructed by broadcast and independent distributed parallel processing in each individual, and a system for introducing the stabilization target in the non-DC range and the control target in the non-DC range appearing on the implementation, and On the way.

In each home, office, etc., in order to cover the power consumption that may occur instantaneously, we contract with the power company for maximum power, and the power company also covers the total power consumption that can be generated by each contract unit I have tried to improve power generation and transmission facilities. However, in summer, for example, the power supply capacity is constantly in crisis. In order to prevent power companies from holding excessive capacity, their supply capacity will remain slightly above demand. As a result, when the demand peaks, the margin becomes extremely small.
In such a group consisting of individuals consuming a large number of resources, the allocation of resources according to the priority of each element while satisfying the restrictions on the total amount of the same resources (power and information transfer capability) Means for doing so dynamically at high speeds while reducing
With respect to so-called static or quasi-static low-speed functions (hereinafter referred to as DC range), the conditions relating to stability are likewise already considered and known.

So far, no means have been provided to analyze the effects of non-DC-like mechanisms placed on targeted power consumers or systems, and to solve them. So far, it has been sufficient to consider the response on the "day" scale, so it has only been considered as a direct current factor. The control of the instantaneous power, even when performed simultaneously, may be affected by various model delays inherent in the system, sometimes causing serious instability. This issue relates to stabilization as a control system for universal objects such as information and energy, as well as resources.

Also, the power flattening control so far has been almost limited to the DC target value, which is the target value of the power regulation value. Foreseeing control that does not deviate from the margin of power, and forecasting demand for power, which is the integral amount of power in a fixed period, and setting the target value by performing large phase operation with control to suppress the protrusion Correspond to. In particular, no means have been provided for achieving both autonomous distributed parallel processing, which is a high-speed control method, and control targets for this non-DC range. This problem similarly relates to setting of control targets for universal objects such as information and energy as well as resources.

The related prior art will be briefly described below.

Japanese Patent Laid-Open Publication No. 11-313438 "Fault Protection Device for Power Distribution System"
In the invention, the hardware is adapted to shut off the circuit in response to the detection of a failure by the failure detector. However, in the present invention, the device group collects information through communication by the server, and It is totally different in that it is necessary to manage the dynamic power coordinated with the required amount in consideration of the priority.

Japanese Patent Laid-Open No. 2001-69668 "Power Management Device"
In the present invention, a method is assumed on which information collection / aggregation is not performed from the device group, but in the present invention, in the device group, the dynamic power management in which the power constraints and the required amount are coordinated in consideration of priority It's totally different from what you do.

JP, 2013-38885, A "in-house power generation system"
Although the invention deals with a power plant, the invention conversely provides a negative power supply.

JP, 2012-85511, A "vehicle charge system which has charge efficiency control and provides an adaptive charge service"
In the present invention, power management in the charging station group is not performed, and the existence of a management station and the existence of a smart grid are assumed. The present invention is completely different in performing coordinated dynamic power management considering power constraints, requirements, and priorities, regardless of the presence of a management station or a smart grid.

JP, 2009-94768, A "power line communication device and automatic registration method of power line communication device"
The present invention relates to a method of establishing a connection in power line communication. The present invention does not specify the communication scheme and does not claim that establishment of communication should be solved. Further, in the present invention, power line communication is listed as one of communication means, but connection establishment there is not listed as a problem to be solved.

JP-A-2004-208393 "Multi-output circuit device capable of setting priority order of power supply"
In the present invention, the excess of the total load current is detected, and it is assumed that the load is disconnected in the set order. In the present invention, the excess load is performed by collecting information in the device group, and a specific detection means is not required. In addition, it is characterized in that the order of cutting the load is not set in advance, but is determined by dynamic judgment in the device group.

JP-A-2003-511842 "Contactor-Breaker"
In the present invention, fault detection and control based thereon are performed in the same individual. The present invention is completely different in that the device group performs dynamic power management coordinated with power constraints and requirements taken into consideration.

JP, 2013-70569, A "distributed power supply system"
In the present invention, failure detection and control are performed in the same individual, but in the present invention, in the device group, dynamic power management coordinated with power constraints, necessary amounts, and priorities taken into consideration. Totally different.

JP, 2011-234561, A "intelligent distribution board, distribution apparatus, blackout countermeasure system, and distribution method"
In the present invention, power failure detection and subsequent connection to a standby power source and reverse operation after power failure recovery are performed in the same individual. The equipment group is completely different in performing coordinated dynamic power management in consideration of power constraints, requirements, and priorities.

JP 2010-148125 A "System for remote acquisition of electrical energy consumption including home use and remote control of distributed target users"
In the present invention, central management based on the communication structure of a central server, a concentrator, and a meter is premised, but in the present invention, in the device group, dynamic power coordinated with consideration of power constraints, necessary amount, and priority. Completely different in that the point of configuration management is to independently manage the management of 的 な in a decentralized local group.

JP 2005-513900 gazette "System for remote acquisition of electrical energy consumption including home use and remote control of distributed target users"
In the present invention, central management based on the communication structure of a central server, a concentrator, and a meter is premised, but in the present invention, in the device group, dynamic power coordinated with consideration of power constraints, necessary amount, and priority. Completely different in that the point of configuration management is to independently manage the management of 的 な in a decentralized local group.

JP, 9-93820, A "solar power generation device"
In the present invention, only communication means and blocking means are described. The present invention is completely different in that the device group performs dynamic power management coordinated with power constraints and requirements taken into consideration.

JP, 10-42481, A "power control device for vehicles"
In the present invention, the power shutoff device is premised on the centralized management configuration forming a tree or star system around it. The present invention is completely different in that the device group performs dynamic power management coordinated with power constraints and requirements taken into consideration.

JP, 2000-16200, A "power control device for vehicles"
In the present invention, the power shutoff device is premised on the centralized management configuration forming a tree or star system around it. The present invention is completely different in that the device group performs dynamic power management coordinated with power constraints and requirements taken into consideration.

JP-A-2005-178778 "Power supply terminal device for automobile and power supply system for automobile"
In the present invention, the power shutoff device is premised on the centralized management configuration forming a tree or star system around it. The present invention is completely different in that the device group performs dynamic power management coordinated with power constraints and requirements taken into consideration.

JP, 2004-348411, A "central supervisory control system integrated type distributed power distribution facility" In the present invention, existence of a central supervisory control system is a premise. The present invention is completely different in that the device group performs dynamic power management coordinated with power constraints and requirements taken into consideration.

JP, 2012-161202, A "tiered supply and demand control device and an electric power system control system"
In the present invention, although a hierarchy is formed, a centralized monitoring control system for collecting information is provided, and it is an embodiment of the present invention, information collection and control within a group are distributed and distributed independently from other groups and hierarchies. There is no point to do, it is totally different.

JP, 2010-279238, A "system monitoring control system"
Although the invention constitutes a hierarchy, it is an embodiment of the present invention in that information collection and control within a group are performed separately and decentrally from other groups and hierarchies, and they are completely different. is there.

JP, 2002-27,686, A "power consumption control method of apparatus in a store"
In the invention, keywords "hierarchy" and "dispersion" appear as names, but as described with reference to the following specification, the contents are different and completely different from the present invention.
Also, each controller constitutes an autonomous distributed system, and even if one controller as a subsystem becomes inoperable, no problem occurs when another controller controls a device under its control. (This is called autonomous controllability), and it is possible to coordinate the purpose of each other between the controllers (this is called autonomous coordination), so that there is a master between each controller. There is no difference between the / slave and the importance, and basically it is possible to carry out management and control with its own resources. "
"Self-supporting decentralized" control in the present invention is, as shown in FIG. 2 of the same specification, the independence between lighting and type controllers such as air conditioners, and power arrangement between the air conditioners under the controller and between the lights. Does not mean that it takes place autonomously. In one embodiment of the present invention, the first feature is that dynamic power allocation is performed independently of other groups or tiers in consideration of priorities among members under them, and the name is And the method is quite different. Furthermore, although “layer” in the invention refers to a hierarchy on the operation mode of time zone, cooperative energy saving, and peak, the hierarchy in the embodiment of the present invention applies to the invention. Is a hierarchy made up of a group of stores as members, or a hierarchy made up of groups of servers made up of servers representing regional stores. The name is “hierarchy”, but the definition is fundamental It is different. In one embodiment of the present invention, it is a first feature that, on any of those layers, dynamic power allocation taking into account the priority between groups is performed independently of other groups and layers, The method is completely different.

JP, 11-45101, A "monitoring control system"
The hierarchy and distributed monitoring control in the present invention are based on decentralization of an execution unit, an information exchange unit, and an interface unit. Dynamically according to the priority among members, under given power constraints, in each hierarchy of which group is a member or in each group of a device as an embodiment of the present invention It is fundamentally different from what provides a way to determine power placement.

Japanese Patent Laid-Open No. 7-308036 "Distribution system monitoring method, distribution system control method and devices therefor"
The supervisory control in the present invention does not implement autonomous control among members. Dynamically according to the priority among members, under given power constraints, in each hierarchy of which group is a member or in each group of a device as an embodiment of the present invention It is fundamentally different from what provides a way to determine power placement.

JP, 7-31013, A "indoor electrical wiring system"
In the invention, only the lighting means of the emergency light is provided.

JP, 2008-90607, A "autonomous distributed type control with resource restrictions"
In the present invention, the term “distribution” means not specifying a server, but the present invention teaches that control is performed in subdivided units so that local and flexible response can be taken, not global collective control. doing. In the present invention, the server may be specified or fixed.
Although the present invention proposes a specific method for performing the measure / decision process, the present invention only needs to define the server by some method, and the method for dynamically allocating the process is not limited. . It does not have to be a card game system, and it does not require special "shift" of server functions. The server function can be either serial number change or fixed.
In the invention, although the purpose is to maintain and achieve the performance by the resource input, in one aspect of the present invention, the power is cut off permanently or intermittently instead of supplying the power as the resource. In this patent, instead of the control for maximizing the supply of total resources, when the power consumption exceeds the allowable value or target value, based on the priority, based on the required power amount and the power amount constraint, the power is permanently stored. Shut off intermittently or intermittently. In the patent, since the power supplied as a resource is determined in advance by the control side, it is information to be measured and acquired in the present invention, while it is known in advance.
That is, the present invention aims to exert "a control function to achieve and maintain the performance of the entire system", and a method of controlling the individual performances of all the elements while satisfying the "total resources (sum of resources)" constraints. (In the same publication, claim 1), in one aspect of the present invention, although there is a restriction on the sum of resources, it is possible to actively sacrifice the achievement and maintenance of performance, The purpose is to prevent damage.

JP, 2013-38470, A "control device and control system of electric equipment"
(1) In the document, constraints of power as resources are not taken into consideration, and a solution that satisfies the constraints is not guaranteed either. It only performs certain operations preset in a feed forward manner. In the claims of the same document, "setting" refers to defining in advance. The present invention is completely different from the invention described in the document in that the present invention explicitly deals with resource constraints and guarantees the operation to satisfy the constraints.
(2) The operation proposed in the same document is the operation of the command by the processing of the broadcast transmission part, but in the present invention, the optimum with constraints is achieved by cooperation between the broadcast transmission processing and the parallel processing in each element. A method of finding the solution of the chemical has been proposed, which is completely different from the invention of the document.
(3) The present invention is characterized in that it is possible to obtain a solution of constrained optimization regardless of dynamic priority change in each element, which is completely different from the invention of the document.

JP, 2011-242030, A "air-conditioning control device"
JP, 2010-19530, A "air conditioning system and communication traffic adjustment method"
JP, 2009-272966, A "equipment management system"
JP, 2007-240084, A "address setting method in air conditioner and air conditioner"
JP, 2007-228234, A "transmission control device, apparatus management system, and transmission control method"
JP, 2004-328184, A "management control system, information transmission method, communication method, network node, transmitting and receiving device, information sharing device, air conditioner and central control device"
These documents only refer to communication address related parts and do not deal with control strategies.

Japanese Patent Application No. 2014-12924 "Power management method and system"
The invention proposes a priority optimization and a power control system that makes it possible in the case where total resources are limited, but these are information collection and allocation from each client of the server It takes the steps of server determination of the amount and notification of quotas from the server to each client, and merely presents the problem to be solved in the present invention. The present invention is intended to speed up processing by sharing the alert element with each member element. The above-mentioned application describing the invention by the present inventor is undisclosed at the time of filing the present application.

Japanese Patent Application No. 2014-153348 "Power Control System, Method, and Information Transmission Capability Control System, Method"
The present invention provides a method of instantaneously allocating resources including power with priority taken into consideration by eliminating parallel communication between server and client, and performing parallel processing of broadcast transmission and independent distribution in each member. ing. However, the present invention only provides a so-called frequency zero or direct current resource allocation method, and response characteristics and information output delay in each member or processing method and hardware in broadcast transmission which occur at the implementation stage. There is no provision at all for instability of the control system due to delays due to wear and stabilization that is the solution. Also, the target to be controlled is the instantaneous resource reception balance, and it only deals with the convergence of the instantaneous power to the regulation value, and similarly, it is carried out by taking the frequency range out of the DC range, preview control or No provision has been made for convergence to the regulation value in consideration of the priority for so-called demand response, which foresees integral control. The invention is therefore quite different from the field providing the solution in the present invention. Although the above-mentioned application itself in which the invention by the present inventor is described is undisclosed, the content is disclosed in International Application WO 2015/115385 and is publicly known.

U.S. Pat. No. 8,504,214 "Self-healing power grid and method thereof"
The disclosure of the document is not related to the power allocation method.

U.S. Pat. No. 8,276,002 "Power delivery in a heterogeneous 3-D stacked apparatus"
Although the document deals with the function of the power supply, it does not deal with the dynamic allocation of power to members.

U.S. Patent No. 8,112,642 "Method and system for controlling power in a chip through a power-performance monitor and control unit"
U.S. Patent No. 7,421,601 "Method and system for controlling power in a chip through a power-performance monitor and control unit"
The document relates to microprocessor power supplies, and not to the function of dynamically and autonomously allocating power.

U.S. Patent No. 7,805,621 "Method and apparatus for providing a bus interface with power management features"
The document discloses power mode transitions, and does not deal with the function of dynamically determining allocation among members.

U.S. Patent No. 6,961,641 "Intra-device communications architecture for managing electrical power distribution and consumption"
The disclosure content of this document is that the intelligent device and the server are configured as a networked network. It has not been disclosed how power management is actually performed. The present invention specifically provides a power allocation strategy.

U.S. Pat. No. 5,581,130 "Circuit board for the control and / or power supply of electrical function devices of a vehicle"
The document only requires that the circuit be modular.

U.S. Patent No. 8,508,540 "Resonant induction to power a graphics processing unit"
The disclosure content of the same document relates to hardware for inductively supplying power, which is completely different from the present invention.

U.S. Pat. No. 8,466,760 "Configurable power supply using MEMS switch"
The disclosure content of this document relates to the switch hardware manufactured by dual substrate MEMS, which is completely different from the present invention.

U.S. Patent No. 7,970,374 "Multi-wideband communications over power lines"
Although the disclosure content of the document relates to transmission media, the present invention does not depend on specific media.

U.S. Patent No. 7,825,325 "Portable lighting and power-generating system"
The disclosure content of the document relates to a specific device and is completely different from the present invention.

U.S. Patent No. 7,755,111 "Programmable power management using a nanotube structure"
The disclosure of the same document relates to a device using nanotubes, which is completely different from the present invention.

U.S. Patent No. 7,320,218 "Method and system for generation of power using stirling engine principles"
The disclosure content of this document relates to hardware called a Stirling engine, which is completely different from the present invention.

U.S. Patent No. 6,965,269 "Microwave phase shifter having an active layer under the phase shifting line and power amplifier using such a phase shifter"
The disclosure content of the document relates to hardware called a phase adjuster in communication equipment, which is completely different from the present invention.

U.S. Patent No. 6,310,439 "Distributed parallel semiconductor device spaced for improved thermal distribution and having reduced power dissipation"
The disclosure content of this document relates to semiconductor placement and thermal diffusion, which is completely different from the present invention.

U.S. Patent No. 6,030,718 "Proton exchange membrane fuel cell power system"
The disclosure content of this document relates to hardware called a fuel cell, which is completely different from the present invention.

U.S. Pat. No. 4,481,774 "Solar canopy and solar augmented wind power station"
The disclosure content of the document relates to an apparatus for solar power generation, which is completely different from the present invention.

"TMC NEWS Hitachi Offers Connected Air Conditioners with Yitran's IT 800 Power Line Communication Chip" Internet <URL: http://technews.tmcnet.com/ivr/news/2005/sep/1186941.htm> or <URL: http: // www.businesswire.com/news/home/20050927005472/en/Hitachi-Offers-Connected-Air-Conditioners-Yitrans-IT800#.UtzSc3xKOSM>
Although the example which attached the communication apparatus to the household appliance is disclosed in the said website, in this invention, it is characterized by performing independent distributed control, without performing communication and centralized control, and the technical content of both is Totally different.

"Smart Home", Internet <URL: http://japan.renesas.com/event/detail/et2011/report/s_home/index.jsp>
The above web site discloses an example of disconnecting a device with a predetermined priority. As compared with the present invention, at least the following points are different.
(1) The example disclosed in the Web site performs control in outlet units, whereas the present invention is in device units and does not limit the use place.
(2) Although the example of the said web site is comprised so that permanent disconnection may be performed, the point which this invention can reduce the electric power not only permanent connection but continuously.
(3) Although the example of the said website performs static priority setting by outlet unit, this invention can perform dynamic priority setting by apparatus unit.
(4) The present invention dynamically determines power consumption allocation within a group, performs independent control among the groups, configures a hierarchy, and dynamically determines power consumption allocation similarly to upper layers. To have.

International application WO 2015/115385 "power control system, method and information transfer capability control system, method"
The contents of the invention are almost the same as the contents of Japanese Patent Application No. 2014-153348. That is, the present invention only provides a so-called frequency zero or DC resource allocation method, and response characteristics and information output delay in each member which occur at the implementation stage, or processing method and hardware in broadcast transmission There is no provision at all for instability of the control system due to delays due to wear and stabilization that is the solution. Also, the target to be controlled is the instantaneous resource reception balance, and it only deals with the convergence of the instantaneous power to the regulation value, and similarly, it is carried out by taking the frequency range out of the DC range, preview control or No provision has been made for convergence to the regulation value in consideration of the priority for so-called demand response, which foresees integral control. The invention is therefore quite different from the field providing the solution in the present invention.

Japanese Patent Application Laid-Open No. 11-313438 JP, 2001-69668, A JP, 2013-38885, A JP 2012-85511 A JP, 2009-94768, A JP 2004-208393 Japanese Patent Publication No. 2003-511842 JP, 2013-70569, A JP 2011-234561 A JP, 2010-148125, A Japanese Patent Application Publication No. 2005-513900 Japanese Patent Application Laid-Open No. 9-93820 Japanese Patent Application Laid-Open No. 10-42481 Japanese Patent Laid-Open No. 2000-16200 Japanese Patent Application Publication No. 2005-178778 JP 2004-348411 A JP, 2012-161202, A JP, 2010-279238, A Japanese Patent Application Laid-Open No. 2002-27686 Japanese Patent Application Laid-Open No. 11-45101 Japanese Patent Application Laid-Open No. 7-308036 Japanese Patent Application Laid-Open No. 7-31013 JP, 2008-90607, A JP, 2013-38470, A JP, 2011-242030, A JP, 2010-19530, A JP, 2009-272966, A JP 2007-240084 A JP 2007-228234 A JP 2004-328184 A U.S. Patent No. 8,588,991 U.S. Patent No. 8,504,214 U.S. Patent No. 8,276,002 U.S. Patent No. 8,112,642 U.S. Patent No. 7,421,601 U.S. Patent No. 7,805,621 U.S. Patent No. 6,961,641 U.S. Patent No. 5,581,130 U.S. Patent No. 8,508,540 U.S. Patent No. 8,466,760 U.S. Patent No. 7,970,374 U.S. Patent No. 7,825,325 U.S. Patent No. 7,755,111 U.S. Patent No. 7,320,218 U.S. Patent No. 6,965,269 U.S. Patent No. 6,310,439 U.S. Patent No. 6,030,718 U.S. Pat. No. 4,481,774 International Application WO 2015/115385

"TMC NEWS Hitachi Offers Connected Air Conditioners with Yitran's IT 800 Power Line Communication Chip", [online], September 27, 2005, Internet <URL: http://technews.tmcnet.com/ivr/news/2005/sep /1186941.htm> or <URL: http://www.businesswire.com/news/home/20050927005472/en/Hitachi-Offers-Connected-Air-Conditioners-Yitrans-IT800#.UtzSc3xKOSM> "Smart Home", [online], Renesas Electronics Corporation, Internet <URL: http://japan.renesas.com/event/detail/et2011/report/s_home/index.jsp>

Unlike the classical centralized control system, the distributed processing type control method that solves the resource allocation policy in consideration of priority has not been studied and studied for stabilization except for the DC region which is a low speed operation area. It is a start. The problem to be solved by the present invention is to introduce stabilization in the non-DC range and a control target in the non-DC range in this distributed processing type control.
The basic principle of the independent distributed control and the stability in the direct current region are well known in the discussion. (Patent Document 49) A group including power consumption elements that participate in independent control, and other power consumption elements that do not necessarily participate in control, including a group (described later) by an independent distributed control method. In addition, the individual who consumes electricity or the individual who switches the electricity supplied to the individuals who consume them permanently or instantaneously repeatedly is called the power consumption factor. Solve the problems that are subject to constraints.
Dynamic and high-speed processing methods using broadcast and independent distributed parallel processing in each individual have already been used as a method of solving resource allocation policy in consideration of priorities while satisfying total resource constraints. It is devised and known. The solution method is clarified also in the past Japanese Patent Application No. 2014-153348, and is disclosed in the international application WO 2015/115385. In the following, the description in the international application WO 2015/115385 is cited, the outline of the method is described as prior art, and it is based on which the description in the present invention is described.

　　　　　　　　　　　　　　　　　　　　　　（１）
ただしｅ^Tはｎ次の単位行ベクトルである（Ｔは転置記号）。 Groups in the domain (Assembly consisting of individuals participating in power control (power consumption elements and participating in control). Defined as the smallest aggregations to be controlled by one broadcast element) Let f ₁ , f ₂ ,..., F _n be power consumption values to be allocated to the respective power consumption elements included in, and let f be a vector in which these are vertically arranged. When the total power regulation value for a group containing a power element and P _t, the constraint that the power consumption total value of the group matches the P _t is expressed by the following equation (1).

(1)
Where e ^T is an n-th unit row vector (T is a transpose).

　　　　　　　　　　　　　　　　　　　　　　（２）
が、上記式（１）の束縛条件の下で極値を取るときのｆ_i（ｉ＝１，２，…ｎ）として、消費電力の割り当て値を求める。なお、上記式（２）中のＱは、対角要素Ｑ_iiがｉ番目の電力消費要素の優先度に等しい正定対称行列である（一般には、対角でなくても正定対称行列であればよいが、以下で優先度を個々に扱う場合は対角として議論できる。説明を簡単にする目的で、以下ではＱをｎ×ｎ対角行列として扱う。）。 Let the current power consumption consumed by each power consumption element in the group be f ^* ₁ , f ^* ₂ ,... F ^* _n, and a vector obtained by arranging these vertically is f ^* . Evaluation function

(2)
The power consumption allocation value is determined as f _i (i = 1, 2,... N) when taking an extremum under the constraint condition of the above equation (1). Note that Q in the above equation (2) is a positive definite symmetric matrix whose diagonal element Q _ii is equal to the priority of the ith power consuming element (generally, it is a positive definite symmetric matrix even if it is not diagonal Although it is good, in the following, priority can be discussed as diagonal if it is treated individually, and in the following, Q will be treated as an n × n diagonal matrix for the sake of simplicity.

　　　　　　　　　　　　　　　　　　　　　　（３）
とすれば（λはラグランジュの未定乗数）、ｆ_i及びλによる上記拡大評価関数の偏微分値がゼロになるという条件からｆ_i及びλの最適解が求められる。 Extended evaluation function

(3)
Then (λ is Lagrange's undetermined multiplier), the optimal solutions of f _i and λ can be obtained from the condition that the partial differential value of the above-mentioned extended evaluation function by f _i and λ becomes zero.

　　　　　　　　　　　　　　　　　　　　　　（４）
　　　　　　　　　　　　　　　　　　　　　　（５）
各個体の優先度の逆数の総和をとり、その総和の逆数を、以下「特性優先度」と呼ぶ。 The optimal solution can be obtained as follows by aggregating the weights (priority) in the group and calculating the partial differential as described above.

(4)
(5)
The reciprocal of the priority of each individual is summed up, and the reciprocal of the sum is hereinafter referred to as "characteristic priority".

The power consumption to be reallocated should be determined as the power closest to the current consumption status of each power consumption element in the group, and therefore the solution is the current power consumption allocation status as in the above equation (5). Depends on

To implement the above power allocation policy, the power consumed by each element in the current group and the priority information of each element are summed up to determine the reassignment policy. Must be notified to each element.
The priority can be dynamically changed depending on the conditions in which each element is placed, such as an allowance for power control at each point in time, the number of people present at the use position, the illuminance or temperature, etc. There is no limitation, but it is grasped and defined in each element. When this operation is performed on a server element provided in the domain, the server first performs an operation to inquire the power consumption and the priority individually for each device element included in the group in the domain, and then After solving the optimization problem, it is necessary to notify or indicate the newly determined and updated allocated power for each device element.
This operation dramatically increases the amount of communication, especially with the increase in the number of elements constituting the group related to power control in the domain, and performs power control at high speed, that is, feedback in real time to limit resources. Make optimization difficult. When the group included in the domain is small and is composed of only 2 to 3 elements, the amount of communication is not so large, but high-speed control is performed with a group composed of hundreds of elements It is difficult to cope with the method in which the server and the client exchange information bi-directionally.
Even if fast response is not required for the control response of the entire domain, if the number of elements constituting the domain is very large, the amount of traffic for communication between servers and clients becomes enormous, making control difficult, and so on Into a difficult situation. When the number of elements constituting a group related to power control in a domain newly emerges or deviates from the group, it is necessary to set the number of elements in the group and parameters related to communication, and the number of elements present and the communication environment It is also necessary to investigate the information, and it is also difficult to be able to cope with the changing situation of the group, which makes it more difficult to perform feedback control in real time.

Japanese Patent Application No. 2014-153348 does not require one-to-one bi-directional communication between the server and the individual clients, and therefore the amount of communication sharply increases even if the number of power consumption elements targeted for power control increases. The power control system and method are provided with excellent scalability because there is no increase and no setting operation for one-to-one communication is necessary. Furthermore, Japanese Patent Application No. 2014-153348 and International Application WO 2015/115385 provide information transmission capability control systems and methods that can be implemented according to the same principle.
That is, it is possible to provide a method of setting the priority in each individual, which can take a substantially constant value without summing the reciprocal of the priority of each individual in the equation (5). There is. This method is a method that can replace the same total with a specified value without aggregation, and communication only needs to broadcast only the excess / deficiency amount with the constraint value of the total resource in equation (5), In the processing in an individual, control is achieved by dividing the information to be broadcasted by the priority set by itself and increasing or decreasing the resource consumption amount in each individual at the present time. High speed processing is possible by broadcast communication and independent distributed parallel processing.

Until now, it was sufficient to consider the response on the "day" scale as control to comply with resource constraints, so it has only been considered as a direct current factor. The control of the instantaneous power, even when performed simultaneously, may be affected by various model delays inherent in the system, sometimes causing serious instability. This issue is common to stabilization as a control system for universal objects such as information and energy as well as resources. An object of the present invention is to provide a means for analyzing the effects of power consumption individuals and non-DC mechanisms placed on the system, using the power system as an application example. The results can be universally applied.

Also, the power flattening control so far has been almost limited to the DC target value, which is the target value of the power regulation value. Foreseeing control that does not deviate from the margin of power, and forecasting demand for power, which is the integral amount of power in a fixed period, and setting the target value by performing large phase operation with control to suppress the protrusion Correspond to. This problem also relates to stabilization as a control system for universal objects such as information and energy as well as resources. An object of the present invention is to provide a solution for a parallel processing method of autonomous distributed which is a high-speed control method and a method for achieving the control target in the non-DC range. The results can be universally applied.

In the present invention, as means for exerting the control function in the non-DC range, (A) “Stability analysis and stabilization means of control system in non-DC range” will be described for continuous system and discrete system. 2.) Two methods are provided: "means for introducing preview control to instantaneous electric power and electric energy that is an integral value of electric power in a fixed section to define a control target in a non-DC region".

First, the "multicast" described in the present invention is defined. A means for delivering information to be shared only in one-direction or one-way information transmission to all individuals in the domain in a significantly shorter time than control time intervals. It is defined as broadcast or broadcast transmission. Information may be delivered with multiple steps depending on the method of configuring the network, but "multicast" or "multicast transmission" described in the present invention does not describe strictly simultaneousness. , Including those cases, will be referred to below as "multicast" or "multicast transmission".

In order to achieve the above object, the measures provided by Japanese Patent Application No. 2014-153348 and International Application WO 2015/115385 are as follows:
The current value of the total power consumed by the broadcast transmission element and one or more individually prioritized power consumption elements, the broadcast transmission element containing in the group one or more power consumption elements And the difference between the total power consumption reference value and the total power consumption adjustment instruction value determined as a function of the difference, to generate information to be shared within the group representing the total power consumption adjustment instruction value, In the group, one or more power consuming elements receive the broadcasted information, and each of the one or more power consuming elements has its own given priority and total power consumption adjustment indication value The power consumption update value to be used for the self power consumption update is determined in parallel independently of the other power consumption elements and the broadcast transmission element among the one or more power consumption elements, Control own power consumption based on power consumption update value More, configured to control the total power consumption in the group, it is a power control system.

In the present invention, as means for exerting the control function in the non-DC range, (A) “Stability analysis and stabilization means of control system in non-DC range” will be described for continuous system and discrete system. 2.) Two methods are provided: "means for introducing preview control to instantaneous electric power and electric energy that is an integral value of electric power in a fixed section to define a control target in a non-DC region".

For this purpose, first, the problems to be dealt with in the present invention are analyzed by a new method.
That is, a control system for allocating resource power that takes into account priority is configured from the control system, forms server-client communication, and independent distributed parallel processing on each individual side that is a multicast transmission and power consumption factor The form to be performed in is clearly identified and analyzed from the description in Japanese Patent Application No. 2014-153348. After that, by describing the phase fluctuation in the system using a dynamic compensator, the characteristics of the system in the non-DC region are analyzed to provide a means for solving the problem.

(A) "Stability analysis and stabilization means of control system in non-direct current region".

(A-1) "New Description of Control System"
(Dynamic model of independent distributed processing in each individual)
The model of power control at each individual side is a first-order lag system for realizing the specified power consumption Pi *, and its block diagram is written as shown in FIG.
(Figure 1 Power control model of each individual)

The closed loop transfer function is
(Fig. 2 Transfer function of each individual)
It is written.

On the other hand, if the adjustment power with the specified power is described as Δ, the expression of the power control system is ((Fig. 3 Control model to the power consumption of each individual for the adjustment amount)
It is.

From the current power consumption Pi, the control system that controls to Pi + Δ is expressed as above. In fact, it is confirmed that the transfer characteristic from Δ to Pi corresponds to the integral of the fluctuation demand Δ.
(Fig. 4 Integration in each individual)

(A-2) "Decision method by centralized processing of feedback gain by server"
(Method of determining feedback gain by server-client two-way communication)
In a power system consisting of a large number of individuals, keeping the total power in the domain constant results in determining a vector as a feedback gain for dividing and distributing the error of the total power, which is a scalar quantity, to the total number of individuals. it can.

The feedback gain can be mathematically calculated from a pole placement problem that determines the responsiveness of the entire control system, or a system control theory that minimizes the evaluation function. In the latter, the driving status of the individual in the domain and the environment information where the individual is placed, such as temperature, illuminance, number of people present, etc. are collected from the individual position where they are obtained, and the weight of the evaluation function, ie In general, it is preferable to set priorities and to calculate appropriate feedback gain vectors.

In order to do so, the server that calculates the same gain vector needs to communicate bi-directionally with the individuals in the domain, which complicates system construction as the number of individuals increases and slows down processing. It was the cause of
(Figure 5 Determination method of feedback gain by server client method)

(A-3) "Determination method of feedback gain by broadcast transmission and independent distributed parallel processing"
(Determination method of feedback gain by broadcast transmission and independent distributed parallel processing)

The operating condition of the individual in the domain and the environment information in which the individual is placed, such as temperature, illuminance, number of people present, etc. are information originally measured and acquired at each individual position. In the invented approach, it is possible to determine the feedback gain vector independently and in parallel, avoiding the collection of priorities in the entire domain. This method is characterized in the method of calculating the priority in each individual, and there is an advantage that the totalized value is properly standardized even if the totaling is not performed in the domain.

Therefore, the server does not have to communicate bi-directionally with the individuals in the domain, and the system uses the total power error information broadcast in the domain and the priority defined in each individual It is possible to proceed with the process by calculating the deviating power amount that each individual should share. As the key is provided as in Japanese Patent Application No. 2014-153348, a mechanism is built in which the characteristic priority of the entire system obtained by totaling them is standardized while setting priorities independently of each individual It is on the point.
(Fig. 6 Determination method of feedback gain by broadcast and independent parallel method (continuous system))

（６）
と書かれ、特に、電力制御機構が同一視できるときは、 

（７）
よって、連続系では、遅れ系がなければ、漸近安定性は、優先度の設定によらずに、自動的に保障される。
  (A-4) "Control system stability (continuous system)"
(Stability of continuous system)
The characteristic equation of the whole domain is

(6)
In particular, when the power control mechanism can be identified,

(7)
Therefore, in the continuous system, as long as there is no delay system, asymptotic stability is automatically ensured regardless of the setting of the priority.

(A-5) "Stability of control system (discrete system)"

（８）
である。
（図７　同報通信と独立並列方式によるフィードバックゲインの決定方式（離散系）） In discrete systems, the stability condition is

(8)
It is.
(Fig. 7 Determination method of feedback gain by broadcast and independent parallel method (discrete system))

（９）
となる。書き換えることで、次式を得、 

（１０）
安定条件として、以下を得る。 

（１１） (Stability of discrete system)
From the block diagram of the discrete system, the characteristic equation of the whole system is

(9)
It becomes. By rewriting, we get

(10)
The following is obtained as the stability condition.

(11)

（１２）
となり、左辺を全系の「特性優先度」として Q* と書けば、 

（１３）
の安定条件が得られる。 (A-6) “The lower limit value of the characteristic priority that gives stability and the convergence time constant of the control system”
(The lower limit of characteristic priority to guarantee stability)
In particular, when the system sensitivity is "1", the stability condition is

(12)
And if the left side is written as Q * as the “characteristic priority” of the whole system,

(13)
The stability condition of is obtained.

In particular, in the case of Q * = 1, convergence is achieved next time, that is, “time constant”, which is an index representative of the time required for convergence described here, is relative to one sample interval, “1 Means to be

（１４）
より、

（１５）
と近似できるので、系全体の収束時定数 τ* は、基準のサンプルインターバル τ に対して、 

（１６）
の関係にあり、無次元時定数を与えていることがわかる。 
  (Converging time constant nondimensionalized at equivalent intervals)
In fact, as a stable condition, there is a lower limit value for the characteristic priority Q * value, but at the same time, its reciprocal means the convergence time constant.

(14)
Than,

(15)
Since the convergence time constant τ * of the whole system can be approximated to

(16)
It can be seen that the dimensionless time constant is given.

(A-7) "Description of phase fluctuation mechanism in control system"
(Individual or stability when there is a delay due to broadcast information development)
There are two major causes of delay in the system.

(Delay in each individual)
(1) Each individual side may not actually measure the power as in the inverter control, but may convert the power consumption from the duty of the PWM and PAM command values (although almost equivalent) . Since this is a command value, control may be performed based on the most recent (delayed) command value issued at the time of broadcast, depending on how to make the logic. In that case, it is equivalent to the presence of a response delay determined by the individual. In fact, the response delay of a high-order system and a simple measurement delay need to be identified. I will mention later.

This delay does not occur if each device correctly refers to the power consumption in the domain that is broadcast at a certain point in time and the power consumption in each individual based on the duty at a very close point. However, when the instantaneous value of the power consumption can be measured only in the minor loop, this delay is often unavoidable.

(Delay of broadcast system)
(2) Delay common to all individuals in the domain. Media-specific broadcast time from sensing module to media converter (media device) to measure total power consumption, and media converter to individual (infrared PPM transmission delay, many hops in Zigbee®) Etc. are common causes for delay. However, depending on the number of hops in the communication media, it usually falls within one interval.

(Uncertainty of response model at each individual side)
Expressively, it can be shown as a measurement time delay of power consumption measured in each individual. (To be exact, it depends on the mechanism of equipment delay. It mentions later.)
In general, it can be described that the transfer function is a mechanism that has some dynamic property, and the result is mediated by the means of F (s).

The delay in the broadcast is usually impossible or very small, unless there is a fair amount of time.
Media hop speed is usually fast enough compared to interval. It is unlikely that the communication delay will be apparent. However, this argument should be regarded as a warning for the vulnerable software defect, as it is possible for the software construction to end up waiting for one interval. As will be described later, a positive response is also possible.
In each individual, for example, in current measurement, there is generally no problem when instantaneous power measurement is performed, but when power measurement is converted by instructions to PWM and PAM, that is, the dynamic characteristics in the local minor loop If you do not consider it, this mistake and misunderstanding are likely to occur.
The present invention provides a solution to how to take measures if a delay is introduced or introduced in error.

Delay in dynamic characteristics or output information of each individual in non-server client type, independent distributed parallel processing type control system, as well as influence on overall system stability caused by transmission delay of broadcast communication system, positive Stabilization methods are hardly studied or discussed. The system of the present invention provides stability evaluation, stability limit conditions, and further, independent distributed system stability by equipping each individual with a condition that “characteristic priority” such as DC gain should have or a dynamic compensator. The method is in fact provided for the first time in the present invention.
Without an understanding of the subject matter of the present invention, stability building is impossible even in a simple system that does not differ in priority from broadcast to broadcast individual. On the other hand, as a feature of the independent distributed parallel control system, since contributions to securing of stability by a plurality of individuals can be expected, securing of stability can be easily realized.
For practical use, the effects of the present invention are extremely large.

(A-8) "Basic measures for system stabilization"
Stabilization includes “reduction of system sensitivity that governs round trip transmission gain of the entire control system”, “compensation of round trip transmission path and phase on each individual”, and “incorporation of phase fluctuation model for each individual There are three means of "dynamic compensation". For this purpose, as described later, Class (class value), which is an index representing each individual's responsiveness and internal delay, is measured in advance by the time of factory shipment, and displayed. It is desirable to use it for coordination when building a control system in a house or office.

(Reduction of system sensitivity)
The easiest and most reliable is the reduction of system sensitivity. This corresponds to increasing the characteristic priority and equivalently increasing the response time constant, and is also a technique for sacrificing the response speed.
Besides operating the system sensitivity directly, it is possible to stabilize the system by intentionally setting high requirements for the “number of individuals in the domain” in the controller, which is a broadcast transmitter, in order to operate the system sensitivity. it can.

(Measurement and display of responsiveness of each individual device (Class class), and adjustment in the implementation system)
In each individual, it is recommended to measure power consumption before reaching the implementation stage, and means for adjustment should be embedded beforehand. The response characteristics obtained on the device side are measured and displayed as "Class value" at the time of factory shipment in advance, and in the system setting stage, the delay information of the entire system is added to the broadcast contents based on the Class value. In the processing on the device side, as described later, it is possible to suppress the delay and achieve stabilization by positively performing the compensation operation.
(Figure 8 Existence of phase fluctuation element such as delay in each individual (continuous system))

（１７）
特に、電力制御機構、遅れ動特性が同一視できるときは、 

（１８）
と記述でき、さらに遅れ系 F(s) を1次遅れ系で近似すると、特性方程式は 

（１９）
と近似できる。  (A-9) "Stability of the system in the non-DC region introducing a phase variation element"
(Evaluation of stability)
The characteristic equation of the entire domain is affected by the lag dynamics.

(17)
In particular, when the power control mechanism and the delay dynamics can be identified,

(18)
If the delay system F (s) is approximated by a first-order delay system, the characteristic equation is

(19)
It can be approximated as

(Impact on stability)
As a result, the characteristic equation becomes a second-order system, the presence of τ causes oscillation, and the presence of the delay system affects the stability of the entire control system.

（２０）
すなわち、特性優先度の逆数 1/Q* 値を小さくとることが有効な条件である。遅れ系のない場合には、離散系でシステムを構成する場合では、上式が「2」以下であることが安定限界であり、「1」以下であることが漸近的な収束性を与えている。遅れ系の存在する場合には、特性方程式が示すように、振動様相を呈する。 (A-10) "Reduction of system sensitivity"
(Low sensitivity)
(1) For stabilization, it is effective to reduce the system sensitivity or its equivalent, and increase the characteristic priority equally.

(20)
That is, it is an effective condition to reduce the reciprocal 1 / Q * value of the characteristic priority. When there is no delay system, when configuring the system as a discrete system, the stability limit is that the above equation is "2" or less, and asymptotic convergence is given that it is "1" or less. There is. In the presence of a lag system, as the characteristic equation shows, it exhibits a vibrational appearance.

In this method, the action is taken by the multicast device. There is no effect on the treatment on each individual.
Manipulating the system sensitivity to be broadcasted or the number of individuals set is one strategy. The stabilization corresponds to setting of intentionally expanding the total number of individuals in the domain or the power adjustable amount in the domain. As a result, the characteristic priority Q * value increases and the response time constant of the system decreases.

(A-11) "Completion of round trip transmission path and phase on each individual"
(Installation and solution of dynamic compensator)
(2) As another means, there is also a policy of providing a dynamic phase compensator in measurement of total power or processing on the equipment side.
In this method, the phase compensator C (s) is built by hardware or software in the broadcast device, or by software in the updating process of the allocated power on the device side, and depending on the delay amount, the phase lead And operate the delay.
(Fig. 9 Phase compensation method in broadcast route (continuous system))

(A-12) "Dynamic compensation method by incorporating phase fluctuation model in each individual"
(Incorporation of phase fluctuation model into processing logic of each individual. Retroactive control)
(3) The destabilizing factor is the power consumption at the time when each individual's power consumption information included in the information of the total power consumption that is broadcasted is referred to / measured by each individual, By acquisition time being different. That is, the phase shift leads to destabilization.
Therefore, in the power control in each individual, it is possible to “match” the information acquisition time by “darely go back to the past” the value of the power consumption referred to perform correction every moment. , This contributes to stabilization. On each individual, a function that can store and save the amount of power consumption referenced and measured in the past will be used.
This countermeasure is made by processing on each individual. There is no impact on the function of the broadcast device.
(Fig. 10 Model-embedded phase compensation method (continuous system) in each individual)

(Extension of stability analysis (continuous system))
In fact, as shown in this block diagram, this correspondence corresponds equivalently to incorporating the delay-causing dynamic characteristic into the power tracking closed loop.
The true purpose of this method is to make the measurement time of the instantaneous power consumption returned / referenced to make the generation time of information measured in the whole domain follow the power command value on each individual It is in. In particular, when the cause of the delay is generated by estimating from the command operation to the device so as to convert from the duty ratio on the individual, the measures mentioned here are each individual rather than going back in the past. It also includes the installation of measures such as aggressive circuits to measure power consumption above.

（２１）
a が十分大きい場合、つまり指令値への電力追従性が高い場合には、近似的に

（２２）
と書かれ、さらに

（２３）
となる。特性方程式は１次系となって、τ の安定性へ与える影響を排除させることができる。応答は、近似的には、τ の大きさに依らず漸近安定的になり、ロバスト安定化がはかれる。ただし、応答速度は、遅れ系における遅れ時間で支配される。 
  The lag system F (s) is approximated by a first-order lag system. The characteristic equation of the whole domain is

(21)
If a is sufficiently large, that is, if the power followability to the command value is high, approximately

(22)
It is written that

(23)
It becomes. The characteristic equation becomes a first-order system, which can eliminate the influence on the stability of τ 2. The response is approximately asymptotically stable regardless of the magnitude of τ and robust stabilization is achieved. However, the response speed is governed by the delay time in the delay system.

(Stabilization by internal embedded model)
This countermeasure will equivalently constitute the control system model of the following figure. In the present invention, this is called a robust stabilization method with an internal embedded model.
(Fig. 11 Equivalent structure of model embedded type phase compensation method in each individual (continuous system))

(A-13) "Response characteristics and delay index Class (class value) in each individual-measurement and display"

（２４）
ここに、指令値の差分を生成する過程が存在すると仮定し、差分量を Δ と表記している。R からΔへの伝達関数は、z変換で記述すると、「(z-1)/z」であり、入力 R に対して「 (π/2 の位相進み) ー (1サンプルインターバル遅れ)」の位相ずれをもつことに注意する。
p は、定常ゲインを補正するために導入している。第1近似では、指令値 R に対する定常ゲインを「1」倍とする場合は、p=m' ととることになる。機器のダイナミクスによっては、定常ゲインは低下する場合もあり、ここでは一般的に p と置いている。p=1 のモデルを「モデル-A」とし、p=m' のモデルを「モデル-B」と呼ぶ。 (A-13-A) "Model-A"
Assuming that the power consumption command value is expressed as R, the dynamics of each individual are

(24)
Here, assuming that there is a process of generating a difference between command values, the difference amount is expressed as Δ. The transfer function from R to Δ is “(z−1) / z” in z conversion, and “(π / 2 phase lead) − (one sample interval delay)” for input R Note that it has a phase shift.
p is introduced to correct the steady state gain. In the first approximation, when the steady-state gain with respect to the command value R is to be “1”, p = m ′. Depending on the dynamics of the instrument, the steady-state gain may decrease and is generally referred to as p here. The model of p = 1 is called "model-A", and the model of p = m 'is called "model-B".

（２５）
Z-変換されたドメインにおいて、その伝達特性は以下で書かれる。 

（２６）
ダイナミクスの高次化に起因する遅れと出力の遅れの2つの効果は、全く性格が異なることに注意する。m' で代表されるダイナミクスの高次化に起因する遅れは、同時に、応答ゲインの低下を招くのに対し、m で代表される出力の遅れは、ゲインの低下を伴わず、位相遅れだけを誘引し、システムの安定性に大きな影響を与える。  The dynamics of each individual and the transient response including its response (output) delay is described below.

(25)
In the Z-transformed domain, its transfer properties are written below.

(26)
It should be noted that the two effects of delay due to higher order of dynamics and delay of output are completely different in character. The delay caused by higher-order dynamics represented by m 'simultaneously causes a decrease in response gain, while the output delay represented by m does not decrease the gain but only the phase delay. Attracts and has a major impact on system stability.

The figure below shows an example of response in a system with m '= 4 and m = 4.
(Fig. 12a Measurement of transfer model of each individual and Class value (discrete system) example a)

(Measurement of class value, class value)
The effect of m 'can be estimated by measuring the amplitude by multiplying the output amplitude of the normal integration process of m' = 1 by "1 / m '".
In the first approximation, for the original input R, including the difference mechanism, there is a delay of “(m ′ + 1) / 2 + m” interval. For the difference amount Δ, it corresponds to the delay of “(m′−1) / 2 + m” interval.

（２７）
Z-変換されたドメインにおいては、その伝達特性は以下で書かれる。 

（２８）
このモデルでは、入力 R に対しては、定常ゲインが「1」に保たれ、位相遅れを生ずるだけの要素にみえる。  (A-13-B) "Model-B"
In the model-A, the integrand is clearly smaller and the steady state gain is reduced. On the other hand, in actual devices, steady-state gain is often maintained, although there is a delay.

(27)
In the Z-transformed domain, its transfer properties are written below.

(28)
In this model, for input R, the steady-state gain is kept at "1", which seems to be the factor that only causes phase lag.

The figure below shows an example of response in a system with m '= 7 and m = 4.
(Fig. 12b Measurement of transfer model of each individual and Class value (discrete system) example b)

The effect of m 'can not be determined from the stationary gain in this model. It may be difficult to measure and estimate separately from the output delay.
In the first approximation, for the original input R, including the difference mechanism, there is a delay of “(m ′ + 1) / 2 + m” interval. For the difference amount Δ, it corresponds to the delay of “(m′−1) / 2 + m” interval.

(A-13-C) "Class (class value)"
Normally, the (m, m ') value can change depending on the "interval time" and "amplitude power of the test input". Therefore, when understanding the device characteristics, both are used as parameters to evaluate the device performance. Need to be They play an extremely important role in the process of adjusting in consideration of the delay characteristic in broadcasting to the domain in the stabilization adjustment of a system composed of devices with many different response characteristics.

(Display Class value)
By performing "Class (m, m ') display" below on the device, stable system settings can be performed.
(Figure 13 Example of displaying the Class value of each individual for the interval and power amplitude)

(A-14) "Use of Class (class value) for stabilization"
This Class value is "adjuster function that compensates for variation of priority definition by almost equalizing convergence time constant different for each individual" and "property priority required to guarantee the stability of the whole control system" It is used by the correction function to be multiplied by degrees.

(Adjustment of Class value (Adjuster) function on the device side)

The value of m greatly affects the stability of the system. Also, the difference in the class value m indicates the difference in responsiveness between devices, and when different individuals (devices) are mixed, the modification function using the characteristics of the priority originally intended is used. It may happen that it is not reflected correctly. This can be corrected by the Class value.

Although it is not desirable to extend the delay by nature, it is difficult to accelerate the response, and aligning the Class m values of the major devices in the system conforms to the intended purpose of the system operation. Have an effect.

(Adjuster function)
It is recommended that each device optionally have an adjuster function to correct this Class m value. For example, in each individual, it is effective to change the delay characteristic of the individual to 2, 4 and so on.
Once the Class values are aligned, broadcast delay is common, and 1) correction of lower limit of characteristic priority, and 2) backward control function on the equipment side can be shared.
That is, the adjuster function refers to the intentional introduction and adjustment of the delay amount for more positively performing stabilization by phase compensation performed in each individual.

(A-14-A) Stability condition in model-A (Figure 14-A Uncertainty model of the whole domain based on Class -A (discrete system))

(Extension of stability analysis (discrete system))
(The stability of the system and the stability condition required for the characteristic priority)

（２９）
と書かれる。 The characteristic equation of the whole system is calculated using l (el) as the number of delay intervals in the same system.

(29)
It is written.

The stability of the system is largely dominated by phase lag.
The total delay, "(l (el) + mi)" is defined as the characteristic delay of the whole system using the maximum value of mi, and will be described again as "l (l)". Also, the order (m'i) representing the dynamics part is described again as "m" below to analyze the stability of the system.

（３０）
 
系全体の特性方程式は、 

（３１）
安定条件は、上式の解が、単位円中にとどまることである。 

（３２）
が単位円を写像した結果の領域が、負の実数 (-1/Q) を含むことが安定の必要条件となる。写像された領域の境界は、z=exp[ωj] とおくと、実軸との交点は、 

（３３）
の解で生ずる。交点の実軸上の値は、 であり、系の特性優先度 Q は安定のための下限値を持ち、その値は 

（３４）
である。特別の場合、l （エル）=0 の場合は、m に依らず、1/2 < Q が安定条件である。 Assuming that the characteristic priority of the system is represented by Q and the system sensitivity is “1”, the analysis can be reduced to the following single equivalent individuals considering the maximum delay.

(30)

The characteristic equation of the whole system is

(31)
The stable condition is that the solution of the above equation stays in the unit circle.

(32)
It is a necessary stability condition that the domain of the result of mapping the unit circle contains negative real number (-1 / Q). The boundary of the mapped region is z = exp [ωj], and the intersection with the real axis is

(33)
It results from the solution of The value on the real axis of the intersection point is, and the system property priority Q has a lower limit value for stability, and its value is

(34)
It is. In the special case, when l (L) = 0, 1⁄2 <Q is a stable condition regardless of m.

It is a feature that stability is guaranteed regardless of the delay of the individual's dynamics if there is no delay of information in the broadcast transmission.

The lower limit for stability of the characteristic priority is as follows. (Here, l (el) = l (el) + m. M = m '.)
(Figure 15-A Lower limit value of characteristic priority for stability based on Class. Model-A)

(Stable limit)
As the dynamics of the device, an increase in order m leading to a decrease in gain results in a decrease in loop gain, so the lower limit of the characteristic priority is reduced and the stability region is expanded. However, it is very important to distinguish whether the delay is due to the dynamics or the output delay, and in the latter case, it has a great influence on the adjustment of stability as shown by the m = 1 column in the above equation. .
As the characteristic priority, since the response vibrates in the values in the above table, it is recommended to take a value of twice or more.

(Improvement based on the class value of stability)
In order to stabilize the system, the reciprocal sum of priority in the domain is normalized to “1”, and for the setting of the number of all individuals (devices) in the domain, at intervals close to actual usage conditions, Corresponding to the Class value measured and displayed in advance with a power adjustment width (both amplitude values at the time of Class test) close to the usage conditions of, it is necessary to correct by multiplying the number of the following individuals. There is. (Here, l (el) = l (el) + m. M = m '.)
(Figure 16-A: Numerical value to be multiplied by the domain individual value required for stabilization based on Class. Model-A)

(A-14-B) Stability condition in model-B (Fig. 14-B Uncertainty model of the whole domain based on Class-B (discrete system))

（３５）
と書かれ、システムの特性優先度を Q で代表させ、システム感度を「1」とすると、モデル-A の場合と同様に、以下の単一等価個体を扱うことに帰着できる。
同様に、全系の特性遅れとして「　l（エル） 」を、ダイナミクス部分の代表次数を「m」と以下記述する。 

（３６）
モデル-A でと同様に、系全体の特性方程式は、 

（３７）
安定条件は、上式の解が、単位円中にとどまることである。

 （３８）
モデル-A での方法を適用すると、系の特性優先度 Q は安定のための下限値を持ち、その値は 

（３９）
である。特別の場合、　l（エル）=0 の場合は、m /2 < Q が安定条件である。
モデル-A では、定常ゲインの低下が安定性確保に働いていたが、モデル-B では、その効果なくなり、安定性に関しては、Q をあまり小さくとれない、すなわち応答を高速化することに限界が与えられる。 
  Model-A Similarly, the characteristic equation of the whole system is

(35)
It can be written that the characteristic priority of the system is represented by Q and the system sensitivity is “1”, which can be reduced to treating the following single equivalent individuals as in the case of Model-A.
Similarly, “l” is described as the characteristic delay of the entire system, and the representative order of the dynamics part is described as “m” below.

(36)
As with Model-A, the characteristic equation of the entire system is

(37)
The stable condition is that the solution of the above equation stays in the unit circle.

(38)
Applying the method of Model-A, the system property priority Q has a lower limit for stability, and its value is

(39)
It is. In the special case, when l = 0, m / 2 <Q is the stable condition.
In Model-A, the decrease in steady-state gain works to ensure stability, but in Model-B, the effect is lost, and in terms of stability, Q can not be made too small, that is, the limit is in speeding up the response. Given.

The lower limit for stability of the characteristic priority is as follows. (l (el) = l (el) + m. m = m '.)
(Figure 15-B Lower limit value of characteristic priority for stability based on Class. Model-B)

In this model-B, where the steady-state gain does not decrease, the characteristic priority becomes higher as m increases. As l increases, the dependence on m gradually declines and takes on a value close to 2 x l (l) / π.
As with model-A, it is recommended that the characteristic priority be at least twice as high as the values in the above table cause the response to oscillate.
Also in Model-B for stabilization of the characteristic priority, it is necessary to multiply the following numbers by the number of devices and correct it. (l (el) = l (el) + m. m = m '.)
As l increases, the dependence on m gradually declines and takes on values close to 4 x l / π.
(Figure 16-B: Numerical value to be multiplied by domain individual value required for stabilization based on Class. Model-B)

Normally, the transmission time delay is overwhelmingly short compared to the broadcast interval and does not pose a problem. However, in the case of broadcasting on a HEMS display server etc., special attention is required as to when the information to be broadcasted is information at that time. If the interval is every 30 minutes, the transmission time can be ignored, but if the content to be broadcast is information 30 minutes ago, the characteristic transmission delay will be one interval, Since the characteristic priority is doubled from the above table, that is, the number of equivalent virtual devices is doubled, the response time constant of the control system is even one hour.

（４０）
モデル－Ｂでは、

（４１）
を安定条件として得る。 
  As a special case, if l (El) is small enough compared to m / 2, then in Model-A

(40)
In Model-B,

(41)
Is obtained as a stable condition.

(A-14-C) "Notes on stabilization"
It should be noted that the representative transmission delay time of the system used here, l (L), is the sum of both "output delay in equipment" and "delay in broadcast system".

In Model-A, what dominates stability is not the dynamics-side order of each individual, but the representative information transmission delay time, and conservatively should refer to the lower limit described in the m = 1 column. is there. If the Class value is correctly obtained at the time of shipment, the number N of each individual (device) in the domain should be “more than twice the value of the table, in order to secure the characteristic priority twice or more of each value of the table. It should be replaced with the number of virtual total individuals (equipment) for stabilization multiplied by the numerical value of.
Model-B instruments are typical, in which case the order also affects the stability, but as l increases, the dependence on the order gradually decreases. If l is small, the effect of the order must also be taken into account.
The former needs to be measured and displayed at the time of factory shipment of the device as described above.
The latter may be due to unrecognizable delays. Factors that are caused by the transmission time itself include the time required for infrared communication and the delay due to multi hop.

The control interval is often determined at the broadcast interval. (In rare cases, such as in the case of decimated reception, this is not the case.) The long interval itself does not lead to control instability. The lower limit value of the above-mentioned characteristic priority is because it is made dimensionless. It should be noted that the increase of the characteristic priority means that the response time constant of the whole system is doubled by the characteristic priority of the control interval, as described above. That is, even if the broadcast interval is shortened to the following, if the characteristic delay time appears, the response time constant of the system reaches a length several to dozens of times the interval.

(A-15) “Differential driving (feedback) and command value driving (feed forward)”

Model that uses the difference amount of power command value as input

The power control function, which takes fluctuation as input, uses the delay m ', and the steady-state gain is stored. For Model-B, it can be expressed as follows.
(Figure 17-1 Differential drive power control)

As described above, in the first approximation, the model -B can be identified as an element having only the phase delay, and can be rewritten as follows. When the difference amount is used as an input, it is written as follows.
(Figure 17-2 Power control with command value drive)

Naturally, the two do not match because they are approximate. However, originally in the model-B, the estimation of m 'is not a problem even if it is identified as the output delay.

Model with power command value as input

The mechanism for extracting the difference amount is “(z−1) / z” in z conversion, and the dynamics for which this is introduced and the power command value is input is model-B.
(Figure 18-1 Transfer characteristic from reference value. Differential drive power control)

However, approximately, as a simple delay mechanism, it can be approximately described as follows.
(Figure 18-2 Transfer characteristic from reference value. Power control with command value drive)

（４２）
であり、最終収束値は、指令値 Pt* に合致する。
この制御の安定性は、特性方程式 

（４３）
で支配される。
システム感度を「1」とした場合、特性優先度に関する安定条件は、 

（４４）
である。
  In control that performs excess power feedback

(42)
The final convergence value matches the command value Pt *.
The stability of this control is a characteristic equation

(43)
Ruled by
When the system sensitivity is “1”, the stability condition regarding the characteristic priority is

(44)
It is.

Approximate model with power command value difference amount as input

In Model-B, stability can be discussed by an approximation model.
(Fig. 19-1 Cycle transfer characteristics from the reference value. Differential drive power control)

Approximate model with power command value as input

（４５）
であり、この所定値を参照する方式では、電力消費は、目標値 Pt* からずれた値に収束する。ただし、多くの使用環境では、これで十分である場合も多い。
この制御の安定性は、特性方程式 

（４６）
で支配される。
システム感度を「1」とした場合、特性優先度に関する安定条件は、 

（４７）
であり、フィードバックを正しく行う場合に比べて、2倍保守的であることを要求する。 
  (Fig. 19-2 Single-cycle transfer characteristic from reference value. Power control with command value drive)
The two are similar at first glance, but the meaning of the amount of input to the control mechanism is completely different.
In the latter case, set m '= (m' + 1) / 2 again.
The final convergence value is

(45)
In the method of referring to this predetermined value, the power consumption converges to a value deviated from the target value Pt *. However, in many usage environments, this is often sufficient.
The stability of this control is a characteristic equation

(46)
Ruled by
When the system sensitivity is “1”, the stability condition regarding the characteristic priority is

(47)
And require that it be twice as conservative as when doing feedback correctly.

Control methods that use power control command values as input should basically be avoided.
If an ammeter is used to measure instantaneous power consumption at all times and used for control, there is no problem with either the convergence value or the stability. The reason to avoid is that the amount of power consumption is not properly reflected in the control, which also causes instability. However, in practice, many cases may occur even when feedforward control is performed only with the command value.
In the inverter control, the instantaneous power is not measured, and the allocated power itself from the power control is used as the duty command value to the device, and the case where the instantaneous power consumption is not measured corresponds to that case. In addition, in the power control of the air conditioner, the case where power consumption is equivalently replaced by the difference between the virtual room temperature and the set temperature corresponds to that.
The description about securing stability here is a discussion on an approximation model for the device performing feedback, and applies to a control system configured with the device performing feed forward.

(A-16) "Dynamic compensation method by incorporating phase fluctuation model in each individual"
(Stabilization method by active introduction of delay or output characteristics to individual processing)

Consider a system consisting of only one individual for simplicity. In the following, F (s) is, for example, a transmission delay, but in general, it may be a dynamic element that manipulates and modifies any output information. They may be elements that produce a PID output.
In control that uses a command value as input, the predetermined value of power is a constant, so it is not necessary to consider it in the analysis of stability. In the following, stabilization will be discussed based on this figure. As described above, since the feedforward control is performed, it affects the control convergence value.
(Fig. 20-1 Model-embedded phase compensation method (discrete system) without compensation in each individual based on Class)

Consider putting F (s) into a power control loop.
(Figure 20-2 Model-based phase compensation (discrete system) compensation based on Class in each individual)

（４８）
と書かれる。 変形すると

（４９）
さらに、

（５０） (Incorporation and stability of the phase fluctuation factor into the processing model of each device)
The characteristic equation that shows the stability of the system is

(48)
It is written. When deformed

(49)
further,

(50)

（５１）
の伝送遅れについては、 

（５２）
となる。第１項は単位円を単位円上に写像するため、安定性確保に関する、特性優先度の下限値への条件は、伝送遅れがない場合に戻すことができ、応答時定数の短縮、高速化につながる。すなわち、(St/Q) < 2 が安定条件となる。 The mapping destination of the unit circle in the z domain can be contained in the mapping destination by the first term of the left side. In special cases, the delay order of the broadcast system is l

(51)
For the transmission delay of

(52)
It becomes. Since the first term maps the unit circle on the unit circle, the condition to the lower limit of the characteristic priority for securing stability can be returned to when there is no transmission delay, shortening the response time constant, speeding up Lead to That is, (St / Q) <2 is a stable condition.

Although the time constant is shortened, it is important to note that it takes time for a high-speed response to be exhibited because the device side pulls out the stored delay information.
(Fig. 21 Model-embedded phase compensation method (discrete system) in each individual)

The compensation method in the processing on this device side can not be processed in common if the dynamic characteristic differs or the output delay differs from device to device, so the characteristic priority lower limit value is fairly uniform as in this example. Can not be improved.

However, it is effective to improve the stability of the entire system, especially by dealing with devices with low priority, ie, high impact, among individual devices.
In a system consisting of at least m '= 1 devices (which is quite common), participating devices share this compensation internally, for delays l common to all systems Is valid. In order to perform this process, the broadcast transmission device side sends l (L) as a common parameter, and the device side that received it executes this.

(A-17) "Implementation of stability evaluation and stabilization, numerical simulation example"
I give a numerical example.

Assuming that the number of individuals is six, the power consumption of each individual is initially set to be as follows. The system configuration of the entire system and the stabilization conditions will be described in detail in "Effects of the Invention" described later.

It is assumed here that the priority in each individual is determined as follows in each individual.
(Fig. 22 Example of domain consisting of six individuals Power unit (W))

In this example, 4000 W is assumed as the regulation value of the total power in the entire domain. That is, in the initial stage, 200 W of power reduction is required.

The system sensitivity shared in the domain starts here as "1".

The total value of the reciprocal of the priority is about 1.4 (0.7 for the characteristic priority), and it is an example where stability can be ensured even if the system sensitivity is “1”, but it exceeds “1”. The response is an example that exhibits oscillatory properties, and is an example that tends to be unstable.
If there is no delay, the transient response is good, and although there is some overshoot, control is complete after 4-5 seconds.
(Fig. 23 Basic response of the system without delay (Characteristic priority = 0.7))

Reducing the system sensitivity has the effect of keeping the vector locus within the unit circle in control of the entire domain and promoting stabilization. The same thing corresponds to increasing the number of individuals in the domain or the amount of power adjustment in the domain to increase the priority in each individual.

The following figure shows an example response when the system sensitivity is 0.3 and when there is no delay system as well. There is no overshoot and it is stable, but it takes 8-9 seconds to achieve control.
(Figure 24 Basic response of system without delay (characteristic priority = 3.3))

Again, return the system sensitivity to "1".
An example of the case of time delay in broadcasting to the domain is shown in the following figure. Although it is described in the broadcast as a delay, in fact, in all individuals, the measurement value of the instantaneous power consumption used for control to follow the power indication value corresponds to a case where one interval is delayed. There is. The effect of one interval delay is significant.

In this example, the delay effect shows a remarkable and violent vibrational response because the tracking controllability in each individual is equivalently high. In fact, the power consumption consumed by each individual does not diverge due to the existence of the upper limit, but the control is practically broken.
(Fig. 25 Basic response of the system when there is a delay of one interval in the broadcast system (Characteristic priority = 0.7))

As mentioned earlier, the most straightforward and easy solution is to reduce system sensitivity.
As described above, the equivalent stability limit at this one-interval delay is "1.0" or less in the priority reciprocal sum value (or "1.0" or more in the reciprocal thereof) in the system of system sensitivity. When adjusting, the stability limit is realized by "0.7" times.
(The characteristic priority corresponding to the stability limit is “1.0”, and the system sensitivity to realize it is “0.7”.)

The following figure shows the response with the system sensitivity set to "0.7". It is understood that “0.7” corresponds to the stability limit, as the theory.
As an extreme example, even in the case where the broadcast delay is “4”, an example where the above-mentioned stability limit can be realized according to theory is also presented.
(Figure 26 Basic response of the system when there is a delay of one interval in the broadcast system (Characteristic priority = 1.0) Example of stability limit.)
(Fig. 27 Basic response of the system when there is a delay of 4 intervals in the broadcast system (Characteristic priority = 2.9) Example of stability limit.)

An example of the response in the case of "0.5" and "0.3" is given below.
Although each has a great effect on stabilization, when the system sensitivity is set to 0.3, the control shows a desirable response in the case where the characteristic priority is further doubled or more as required above. There is. However, it takes about 13 seconds to complete the control.
(Fig. 28 Basic response of the system when there is a 4 interval delay in the broadcast system (characteristic priority = 5.8))
(Fig. 29 Basic response of the system when there is a delay of 4 intervals in the broadcast system (Characteristic priority = 9.7))

The example of response which introduced the phase compensator in the loop which feeds back the total electric energy is shown.
The change between intervals of the total electric energy measured in the domain is measured and multiplied by an appropriate coefficient to be mixed and broadcast to the total electric power shortage and surplus information. The transfer function of the phase compensator is typically “C (s) = (1 + ks)”
It is written.
Since the priority of each device in the domain varies widely, the improvement by one dynamic compensator is not effective and the stability is improved but the effect is limited.
(Fig. 30 Example of response with phase compensation element in broadcast system (Characteristic priority = 0.7))

In each device in the domain, a response example is shown in which the value one interval earlier stored in each device is retroactively used as the instantaneous power consumption to be fed back and referenced in the following control.
The effect is very effective as expected, and although it takes one interval to read the past, control is completed for about 9 to 10 seconds. The system sensitivity is not reduced at "1".

An example of improvement in this retroactive compensation is also shown for the case of the broadcast delay being "4". Again, the system sensitivity is not reduced at "1".
(Fig. 31 Example of response when the delay model is incorporated, where all individuals go back one interval when the broadcast delay is 1 interval) (Characteristic priority = 0.7))
(Fig. 32 Example of response at the time of delay model incorporation, with all individuals going back four intervals when the broadcast delay is 4 intervals, (characteristic priority = 0.7))

In this example, since it is an example in which a time delay is assumed for the transmission of the broadcast value to the whole system, the measures in all individuals have been effective, but in practice, it is possible to identify the part causing the delay. There are also difficult cases.
One possible countermeasure is to refer to stored past information as information to be returned / referenced in follow-up control in the main consuming individual in the domain. This is an example in which the main individual tries to compensate for the delay in the whole system.
The following response example is an example when this measure is taken only for individual -1, -6. Certainly, the improvement of stability has come to an end. However, it can not be said that it is enough. Here, the system sensitivity remains "1".

As to how to take measures, as described later, it is effective to use measures to reduce the system sensitivity in combination, which is a method of maintaining responsiveness.
(Fig. 33 An example of response when the delay model is incorporated, where the major delay only occurs one interval back when the broadcast delay is 1 interval (characteristic priority = 0.7))

In the following example, there will be described a case where there is one broadcast delay in the domain, and in the individuals -1 and -6, further, there is a device-specific delay of one interval.
The error power history is in a very oscillatory and unstable state. Also here, the system sensitivity is intentionally fixed at "1".
(Fig. 34 Response example when there is a delay of one more interval for two individuals when the broadcast delay is one interval) (Characteristic priority = 0.7) Instability.

First, the results in the case where retrospective control to the past is performed only for the multicast delay in each individual in the domain are listed. That is, it corresponds to the case of not dealing with the delay for each individual occurring uniquely in the individual -1, -6. Here, the system sensitivity remains "1".

The effect is clear and effective. However, control is still not complete after 15 seconds.
(Fig. 35 Response example in which retroactive control is performed for one interval for all individuals when there is a delay of one interval for two individuals when the broadcast delay is one interval) (Characteristic priority = 0.7)

Subsequently, for each individual in the domain, the results for the case where retrospective control is performed, in which the corresponding ones and individual delays are also present and summed up, are listed. Here, the system sensitivity remains "1".

The effect is further improved, and it can be seen that the control is completed in about 13 seconds.
(Fig. 36 A response example in which, when the broadcast delay is 1 interval, there is a delay of 1 interval with 2 individuals, and the individual side adds up and compensates each retroactive control (characteristic priority = 0.7))

Again, the ultimate stabilization method is a combination of measures to reduce system sensitivity.
First, the system sensitivity is set to 0.3, and the response when no retroactive control is performed is listed.

If the broadcast delay is “1” and the output delay for all devices is “1”, the system sensitivity corresponding to the stability limit is 0.42, and 0.3 corresponds to an example in which the stability is not sufficient. .

As expected, the stabilization has a great effect, but the completion of the control is well ahead, and the long period oscillation mode continues and is not good.
(Fig. 37 In the case where the broadcast delay is 1 interval and there is a delay of 1 interval in 2 individuals, a response example in which only the system sensitivity is reduced without compensation on the individual side (characteristic priority = 2.4))

The final example is the case of combining system sensitivity reduction measures with retrospective control in each individual.
It is understood that the response is improved as well as the stability is improved, the overshoot is completed, and the control is completed in about 13 seconds.

The last set of measures mentioned is the final adjustment means.
Correctly, from the measurement and processing means of the instantaneous electric energy that is returned / referenced in the power tracking control on each individual, excluding the delay inherent in the measuring means and broadcasting means of the total power in information processing Elimination of the cause of delay is a priority to be taken.
The means described here should be adopted as the final adjustment means in an environment where the measures can not be thoroughly implemented.
(Fig. 38 A response example in which, when there is a delay of 1 interval for two broadcasts with a delay of 1 interval, each individual retroactive control is added and compensated on the individual side, and the system sensitivity is reduced (characteristic priority) Degree = 2.4))

(B) About "means for introducing preview control to instantaneous electric power and electric energy that is an integral value of electric power in a fixed section and defining a control target in a non-DC region".

(B-1) "Perspective control of instantaneous power"
(Foresight control on instantaneous power and avoidance of overshoot)
On each individual, the maximum total electric energy that can be input on individual individuals by referring to the shortfall / remaining amount of total power consumption within the instantaneous domain that is broadcasted You can know Precisely, by sharing the surplus power amount according to the priority, it becomes possible to perform an operation to prevent a plurality of individuals from simultaneously consuming the surplus power amount and generating an overshoot.

This method is a kind of phase operation and is to increase power consumption on an individual from now on, that is, resources in the future using future electric power consumption in each individual in consideration of the advance of the phase due to differential operation. Is equivalent to executing the assignment of
The actual operation can not be expressed by a simple linear transfer function, but as shown on the next page, PD or PID operation is performed by performing proportional operation (P operation) in which each individual is multiplied by the reciprocal of priority. You may think that it will be achieved.

(B-2) "Integrated power, preview control to demand"
In the sense of controlling the integral quantity, there is also a policy of introducing a PID control compensator in the broadcast loop. H0 (s) in the figure corresponds to that. Instead of pure integration guarantee, it may be a mechanism that outputs a moving average for the past one month, in which case the monthly power demand will be fed back.

“Demand” is a term that indicates the amount of integrated power within a fixed period, but it is the output of a low-pass filter with a so-called large time constant, and it is a phase converter placed on the broadcast path. It is a control target value in the non-DC range that can be obtained.
In terms of control, it is also possible to control this demand as a resource, which can also be considered as part of a stabilization means for arranging a phase compensator on the above mentioned broadcast route, but here the demand is preview control It provides a control method to place resource allocation in advance. The concept of the prior arrangement is a new concept provided by the present invention in that it is implemented in consideration of priority in parallel processing of broadcast transmission and independent distribution.

In the figure, Hi (s) indicates a function of phase operation such as PID in linear operation, or a phase operation in a broad sense that generally utilizes nonlinearity.
(Fig. 39 System configuration which introduced respective compensators on both the broadcast system and individual side)

(B-3) "Example of preview control of instantaneous power"
(Example of preview control of supply current in substation pipe)
(1) An example is given in which the current flowing out of the substation is restricted in the case where the train is controlled to repeat acceleration, additional notches, inertial operation, etc. The substation supplies current to a plurality of trains in the pipe. In a certain train, acceleration is performed according to the schedule, but when the current supplied from the substation is regulated by 750A. In this example, the current at the maximum acceleration of the train is 2000A, and only P operation and without introducing the preview function, it is not possible to enter peak cut control without stepping once over 750A.
(Fig. 40 In a system consisting of a railway car and a substation, an example of a control result in which occurrence of overshoot is suppressed by previewing and controlling the current sent from the substation)

Overshooting can be avoided by the train loading logic sequentially adding an additional current value to be assigned to itself based on the value of the reported extra current.

(B-4) "Integrated power, example of preview control to demand"
(Phase leading operation of deviation power history (peak shift))

In so-called proportional operation (P 2 operation), suppression control of power consumption is started after deviation starts, that is, with overshoot. The effect is to peak cut the power at the expense of a rise in temperature in the maintenance room or in the refrigeration showcase as a result. In this method, the temperature inside and in the storage room causes the target temperature to rise, and overshoots are inevitable.

Avoiding the temperature rise in the room and the storage room by keeping the total integrated electric energy constant (consequently as I operation) by consuming the reduction of the deviating power in advance of the deviating amount in the future (and as a result I power) Can be flattened. This operation is an example in which the effect can be more effectively exhibited in PID control that performs phase lead compensation control (D operation) in a broad sense.

(Peak shift by battery and refrigerator)
(Fig. 41 The concept of control that predicts the amount of energy deviating from the peak and shifts it in front of it)

(Place the estimated amount of total power deviation ahead of the time zone in advance.)
(Fig. 42 The concept of control that predicts the deviating power amount from the peak and shifts it in front of it. Transition of the amount of power adjustment.)

(As a result, the total deviation of the deviation period is replaced with the total amount of power to be arranged in advance.)
(Fig. 43 Power history obtained as a result of peak shift control.)

The resulting concept of peak-cut power history.

(Meaning of phase compensation control of deviant power history)

An air conditioner in a room having a refrigerator, or a system having a refrigeration showcase, as a result, accommodates storing the future deviating electric energy by reducing the temperature in the room and in the room by them. In the case of having a battery, future deviating power will be stored as power in advance. In practice, it is effective to combine both the battery and the indoor / interior temperature reduction and compensate the suppression control of the additional deviating power due to the weather prediction error in the P 2 operation. These will achieve broad phase compensation control combining PID operation.

(Figure 44 Temperature history in the cabinet obtained as a result of peak shift control.)

The total amount of deviating power is accumulated as the temperature decrease in the freezer showcase.

"Processing algorithm"
In order to exhibit the function in the non-DC range, using dynamic compensators, allocate resource power with priority taken into account by independent distributed parallel processing on each individual side that is broadcasting and power consumption elements The algorithm is illustrated.
(Fig. 45 Processing flow for incorporating and implementing dynamic compensation for performing phase correction and the like on broadcast transmission elements and power consumption elements)

In the conventional IP, only P operation, which is a DC gain, is assumed in the broadcast transmission element and the power consumption element.

By incorporating a dynamic compensator into this, it is possible to realize the functions of the present invention such as phase compensation control, preview control, peak shift control and the like including PID operation.

This figure shows the algorithm.

"Example of assumed device"

The device to be realized (1a) is also called a sensing module.
(Fig. 46 Example of a functional block diagram of a device that measures the current, incorporates a phase compensator, and broadcasts it)

In conventional IP, as information generated by the broadcast transmission element, only the P 2 operation, which is a DC gain, is assumed. The present invention can be realized by providing a dynamic compensator including phase operation and the like including PID operation.

The device to be realized (1b) makes a fatal contribution in this scheme.
(Fig. 47 An example of a functional block diagram of a device that secures driving power by induced current, measures current, and incorporates a phase compensator to broadcast simultaneously)

Device to be realized (2) The simplest way to reduce the power is to shield and restore the stepless or intermittent circuit.
(Fig. 48 An example of a functional block diagram of a device that incorporates a phase compensator and shuts down the circuit to control the duty in a power consuming individual)

In conventional intellectual property, only P operation, which is DC gain, is assumed in the power consumption factor. The present invention can be realized by providing a dynamic compensator including phase operation and the like including PID operation.

In the device to be realized (3), although the circuit is not shut off and restored explicitly, there may be a device that generates an inverter drive signal to control the duty of power consumption.
(FIG. 49 shows an example of a functional block diagram of a device that incorporates a phase compensator and has a signal generation mechanism that adjusts the duty to shut off the circuit in a power consuming individual)

Similarly, even in the case where the power consumption element has a function to perform intermittent switching, the present invention can be realized by newly providing a dynamic compensator including phase operation including PID operation. .

The properties of the characteristic response of the system response by the introduction of the stability, the stability limit, and the dynamic compensator will be described in further detail by delving into the example of Example 1. In the present example, the details regarding quantitative stability are also described.
In a system that performs control by independent distributed parallel processing in the same report and each individual, there are many cases where delay in the delivery of the same can not be avoided even when there is no delay in the individual side processing process. This occurs because the measurement facility of total resources such as total electric power, the device is a means of low speed, or measurement is performed at a facility or device several stages ahead. It is necessary that the consumption of resources such as the power consumed by each individual be reflected in the amount of consumption that is broadcast, that is, the coincidence is necessary, and if this is not satisfied, as described above , Need to improve stability. For stabilization, as summarized in the means to be solved, the characteristic priority of the control system is raised above the stability limit value to slow down the response of the control system, or the delay of the system is taken in as an internal model at each individual side. There is a way to dynamically compensate at. The latter is a structure in which system stabilization is performed by parallel processing of independent distributed as in resource allocation.

In the configuration of Example 1, consider the case where there is no delay on the individual side. The broadcast delay is assumed to be four intervals. As described in the solution means, the stability limit in this case is the characteristic priority of 2.9 (explained in FIG. 27), and is unstable unless it exceeds this.
(Fig. 50 Instability without compensation. (Characteristic priority = 0.7))
The figure shows that the characteristic priority is unstable at 0.7 with no compensation at all.

If the property priority is increased to 7.1, the stability can be ensured, as theoretically.
(Fig. 51 When characteristic priority = 7.1, no compensation is made on the individual side)
But the response is very slow.

In theory, the control should be completed at the next interval, the system sensitivity is adjusted to give a characteristic priority of 1.0, and the results of compensation that each individual incorporates a four-interval delay model are shown below. Shown in.
(When the characteristic priority is set to 1.0 in FIG. 52 and compensation is performed with an internal model equivalent to four intervals in each individual)
Although the delay itself remains, the response is performed instantaneously according to the theory and can exhibit excellent responsiveness. This indicates that the independent distributed parallel processing is performed to exhibit the structure for securing the stability of the entire system.

However, care must be taken in using this dynamic compensation method. FIG. 53 shows the case where each individual sets the delay compensation incorrectly as 5 intervals, and FIG. 54 shows the response when the broadcast delay is incorrectly set as 3 intervals. In fact, the limiter does not work so much because the limiter works, but it shows that the control is broken. In order to incorporate delay into a model, broadcast delay must be managed. In practice, this management requirement is not severe, and it is a positive response method to intentionally increase the delay and remove the uncertainty. As a result, responsiveness improves dramatically.
(Fig. 53 With the characteristic priority = 1.0, and with the broadcast delay being 4 intervals, each individual performs compensation incorporating an internal model equivalent to 5 intervals)
(Fig. 54 When characteristic priority = 1.0 and the broadcast delay is 4 intervals, each individual performs compensation including an internal model equivalent to 3 intervals)
When each individual has uniform characteristics, the amount of delay to be dealt with is the sum of the broadcast delay and the delay of one type of individual, and the response method is clear. However, in a system composed of various devices such as home power control, the delay amount differs for each individual, and it is not easy to cope with it.

In the following, we will describe representative measures for them.
In the configuration of the embodiment 1, the case where there are 5 in the individual 1 and 2 intervals in the individual 6 is considered. Fig. 55 shows the response that attempted to stabilize only with the characteristic priority when the characteristic priority was set to 7.1 (system sensitivity = 0.1) and there was no individual-side compensation when there was a broadcast delay of one interval. Listed.
(Fig. 55 With property priority = 7.1, no compensation on the individual side)
The response is quite slow.

Similar results were described in Example 1. However, in the case of only the main individuals, the responses dealt with the sum of the broadcast delay and the delay for each individual are shown in FIG. Since the delay in the individual varies, it can be seen that although the response is not uniform, it can be converged relatively fast.
(Fig. 56: Characteristic priority = 1.78, and the same report and compensation for individual delay are performed on the main individual side)
As the characteristic priority is low, the response is faster.

Thus, when the delay in each individual is different, it may be better to positively unify the delay. If the uncertainty of the model is removed, and even if the delay is large as a result, if the compensation by incorporating the internal model is provided, the characteristic priority can be reduced to the contrary and the response can be speeded up. In FIG. 57, the individual side delay is intentionally increased to be uniform according to the device with the maximum delay on the individual side, and an internal model corresponding to the total delay amount obtained by adding the broadcast delay time is incorporated. It is a stabilized example. Here, it can be seen that the control at the ideal characteristic priority of 1.0 is completely implemented. The method of removing uncertainty while increasing the delay is a good example of producing a fast response.
(Fig. 57 With the characteristic priority = 1.0, the individual delay amount is unified on all individuals side, the broadcast delay is added, and the internal model built-in compensation for 6 interval delay is performed)

The addition of the delay amount to remove the uncertainty should be made proactively. Although there is also a way of thinking performed only by the main individual, in the case of the compensation method with reference to the past, the error and uncertainty may be noticeable. The result shown in FIG. 58 is the result of performing the method in FIG. 57 on only some of the main devices, and although the control is functioning, an overshoot occurs and an oscillatory property appears.
(Fig. 58 With the characteristic priority = 1.0, the individual delay is unified only on the main individual side, the broadcast delay is added, and the internal model built-in compensation for 6 interval delay is performed)

As described above, the function of adjusting the amount of delay is expected for the next generation of independent distributed parallel control, by apparently delaying the delay on the individual side to remove uncertainty. This concept is also applied to the transmission of broadcasts, and a function to be a intentionally managed delay is similarly required in the next generation of independent distributed parallel control.

（５３）
ここに

（５４）
である。近似的に

（５５）
が成立し、

（５６）
の特性優先度の定義を用いて

（５７）
のように、遅れの優先度の逆数の重み付き平均、特性遅れ量　y* を用いることで、以下の安定化条件を得る。

（５８）
書き換えると、m が共通の場合には

（５９）
となり、さらに m=0 の場合には

（６０）
が安定条件である。
内部応答モデルを正しく用いた場合の、典型的な内部応答モデルを図５９に、また、それを総合したモデルを図６０に、全系のモデル例を図６１に掲げた。
（図５９　典型的な内部応答モデル）
（図６０　内部応答モデルの総合モデル）
（図６１　全系モデルの例） (Evaluation of stability)
(I) When the delay model is not internally incorporated Consider a system combining the internal output delay mi and the notification delay I. Under the simplest system structure, when the system sensitivity is "1", the characteristic equation is as follows.

(53)
here

(54)
It is. Approximately

(55)
Is established,

(56)
Using the definition of the characteristic priority of

(57)
The following stabilization conditions are obtained by using the weighted average of the reciprocal of the priority of the delay and the characteristic delay amount y *.

(58)
Rewriting, if m is common,

(59)
And if m = 0, then

(60)
Is the stable condition.
A typical internal response model when the internal response model is correctly used is shown in FIG. 59, a model obtained by combining it is shown in FIG. 60, and a model example of the whole system is shown in FIG.
(Fig. 59 Typical internal response model)
(Fig. 60 Integrated model of internal response model)
(Fig. 61 Example of whole system model)

（６１）
近似的には、
 

（６２）
と等価でもある。こうすることで m'は陽にでなくてよいとわかる。遅れと応答は

（６３）
で記述することができる。もし、yi が全個体に共通ならば、y に寄らずに安定条件は、

（６４）
となる。これは非常に有用な近似安定条件である。mi' に合わせて、積極的に mi  を操作して、yi を共通にする意義が認められる。 The stability of the system is determined by the following characteristic equation:

(61)
Approximately

(62)
It is also equivalent to It turns out that m 'does not have to be positive. Delay and response

(63)
Can be described by If yi is common to all individuals, the stable condition is

(64)
It becomes. This is a very useful approximate stability condition. The significance of making yi common is recognized by positively manipulating mi according to mi '.

(II) Incorporating a delay model internally Stabilizing operating characteristic sensitivity or system sensitivity must be adjusting DC gain. As described above, in each individual, the stabilization conditions in the stabilization method in which the delay model is incorporated are generally described.
A typical model incorporating an internal model is shown in FIG. 62, a model combining the models is shown in FIG. 63, and an example of the entire system is shown in FIG.
(Fig. 62 Typical model incorporating internal model)
(Fig. 63 Comprehensive model incorporating internal model)
(Fig. 64 Example of whole system model incorporating internal model)

（６５）
ここに、xi は

（６６）
である。Δxi を個体側で意図的に設定させることで、高速の応答を維持できる制御を行うことができる。各個体側でのモデルは、

（６７）
の近似により、以下の特性方程式を得る。

（６８）
ここに

（６９）
の関係があるため、結局は次式が安定条件となり、理想的な安定性が約束される。

（７０）
さらに、xi が共通の場合には、近似を経ずに、厳密な取り扱いも可能になる。

（７１） The stability of the whole system is governed by the position of the roots of the following characteristic equation:

(65)
Where xi is

(66)
It is. By setting Δxi intentionally on the individual side, it is possible to perform control that can maintain a high-speed response. The model at each individual side is

(67)
The following characteristic equation is obtained by approximation of

(68)
here

(69)
In the end, the following equation is a stable condition, and an ideal stability is promised.

(70)
Furthermore, if xi is common, exact handling is possible without approximation.

(71)

It has to be noted that a reduction of the characteristic frequency, i.e. an increase of the natural period takes place, as can be seen in this strict handling. In the case of m '= 1 often appearing in practice, the output delay and the broadcast delay have the following properties, as indicated by l (el).
Here, let ZI be the characteristic root when not incorporating the delay model inside, and let Z-II be the characteristic root when incorporating the delay model inside. (* Analysis by the following ZI is approximate.)

At the stability limit, it is understood that the root is -1 for both ZI and Z-II, and the stability limit is Q * -I = 1/2 + 1 (El), Q * -II = 1/2.
Under practical conditions, Q * -I = 1 + 2 * l (El), Q * -II = 1 should be taken on both sides, and in this case, the corresponding characteristic root is ZI = l (El) / (l (el) +1), Z−II = 0. This Z-II value indicates an ideal response.

For ZI solutions: Although approximately, Q * -I = 1 + l (El), and ZI = 0. However, the analysis here appears to be generated because it is approximate, and although the result is attractive, in practice, such excellent properties can not be expected.
About the Z-II solution: The analysis result is strict, and with the correct setting, the ideal response of Z-II = 0 is obtained.

Power control model of each individual Transfer function of each individual Control model for power consumption of each individual for adjustment amount Integration in each individual Determination method of feedback gain by server client method Determination method of feedback gain by broadcast communication and independent parallel method (continuous system) Determination method of feedback gain by broadcast communication and independent parallel method (discrete system) Existence of phase fluctuation element such as delay in each individual (continuous system) Phase compensation method in broadcast route (continuous system) Model embedded phase compensation method (continuous system) in each individual Equivalent structure (continuous system) of model embedded type phase compensation method in each individual Measurement of transfer model of each individual and Class value (discrete system) example a Measurement of transfer model of each individual and Class value (discrete system) example b Example of displaying Class value of each individual for interval and power amplitude Domain-wide uncertainty model-A (discrete system) based on Class Domain-wide uncertainty model-B (discrete system) based on Class Lower limit of characteristic priority for stability based on Class. Model-A Lower limit of characteristic priority for stability based on Class. Model-B A numerical value based on Class that should be multiplied by the individual domain value required for stabilization. Model-A A numerical value based on Class that should be multiplied by the individual domain value required for stabilization. Model-B Differential drive power control Command value driven power control Transfer characteristics from the reference value. Differential drive power control Transfer characteristics from the reference value. Command value driven power control Transfer characteristics from the reference value. Differential drive power control Transfer characteristics from the reference value. Command value driven power control Model-embedded phase compensation method (discrete system) without compensation for each individual based on Class Model-embedded phase compensation method (discrete system) compensation for each individual based on Class Model embedded phase compensation method (discrete system) in each individual Example of a domain of six individuals Power unit (W) Basic response of the system without delay (characteristic priority = 0.7) Basic response of the system without delay (characteristic priority = 3.3) Basic response of the system when there is a delay of one interval in the broadcast system (characteristic priority = 0.7) Basic response of the system when there is a delay of one interval in the broadcast system (Characteristic priority = 1.0) Example of the stability limit. Basic response of the system when there is a delay of 4 intervals in the broadcast system (Characteristic priority = 2.9) Example of stability limit. Basic response of the system when there is a delay of 4 intervals in the broadcast system (characteristic priority = 5.8) Basic response of the system when there is a 4 interval delay in the broadcast system (characteristic priority = 9.7) Example of response in which phase compensation element is added to the broadcast system (characteristic priority = 0.7) Example of response at the time of delay model built-in, where the broadcast delay is 1 interval, and all individuals go back by 1 interval (characteristic priority = 0.7) Example of response when the delay model is built in, where the broadcast delay is 4 intervals in all the individuals when the delay is 4 intervals (characteristic priority = 0.7) Example of response at the time of delay model built-in, in which the delay is 1 interval, and the main individual only goes back by 1 interval (characteristic priority = 0.7) Example of response when there is a delay of 1 interval for 2 individuals when the broadcast delay is 1 interval (characteristic priority = 0.7) Unstable. Example of response in which retroactive control is performed for one interval in all individuals when there is a delay of one interval in two individuals when the broadcast delay is one interval, and the characteristic is the case (characteristic priority = 0.7) In the case where there is a delay of one interval in two individuals when the broadcast delay is one interval, a response example in which individual retrospective control is added and compensated on the individual side (characteristic priority = 0.7) In the case where there is a delay of one interval in two individuals when the broadcast delay is one interval, an example of response in which only the system sensitivity is reduced without compensation on the individual side (characteristic priority = 2.4) An example of response where the retrospective control is added together and compensated on the individual side and the system sensitivity is reduced when there is a delay of 1 interval between two individuals when the broadcast delay is 1 interval, and the system sensitivity is reduced (characteristic priority = 2.4 ) System configuration which introduced each compensator on both the broadcast and individual side In a system consisting of a railway car and a substation, an example of a control result in which occurrence of overshoot is suppressed by previewing and controlling the current sent from the substation The concept of control that predicts the amount of energy deviating from the peak and shifts it in front of it The concept of control that predicts the amount of power deviating from the peak and shifts it in front of it. Transition of power adjustment amount. Power history resulting from peak shift control. Internal temperature history obtained as a result of peak shift control. Processing flow that incorporates and implements dynamic compensation that performs phase correction, etc. on broadcast transmission elements and power consumption elements Example of a functional block diagram of a device that measures current and incorporates a phase compensator to broadcast simultaneously Example of a functional block diagram of a device that measures driving current by inductive current, measures current, and incorporates a phase compensator to broadcast An example of a functional block diagram of a device that incorporates a phase compensator and shuts down the circuit to control the duty in a power consuming individual An example of a functional block diagram of a device that incorporates a phase compensator and has a signal generation mechanism that adjusts a duty to shut off a circuit in a power consuming individual Instability without compensation. (Characteristic priority = 0.7) When characteristic priority = 7.1 and no compensation is made on the individual side When the characteristic priority is set to 1.0, and each individual performs compensation incorporating an internal model equivalent to 4 intervals When the characteristic priority is set to 1.0, and the broadcast delay is 4 intervals, each individual performs compensation including an internal model equivalent to 5 intervals When the characteristic priority is set to 1.0, and the broadcast delay is 4 intervals, each individual performs compensation including an internal model equivalent to 3 intervals. When characteristic priority = 7.1 and compensation on the individual side is not performed When characteristic priority = 1.78 and the same report and compensation for individual delay are performed on the main individual side When characteristic priority is set to 1.0, the individual delay amount is unified on all individuals side, the broadcast delay is added, and the internal model built-in compensation for 6 interval delay is performed When characteristic priority = 1.0, the individual delay amount is unified only on the main individual side, the broadcast delay is added, and the internal model built-in compensation for 6 interval delay is performed Typical internal response model Integrated model of internal response model Example of whole system model Typical model incorporating an internal model Comprehensive model incorporating internal model Example of whole system model incorporating internal model An example of a data processing apparatus which performs high speed and stable processing by managing broadcast transmission and delay of individual side generated data Application example in water supply

The present invention is a stabilization system and method of a control method including broadcast transmission and independent distributed parallel processing, which uses power as a resource and allocates resources in consideration of priority when total resources are restricted. To solve. A resource allocation method in a low speed direct current region is solved in Japanese Patent Application No. 2014-153348 and International Application WO 2015/115385, and the latter is already known.

The present invention can be widely applied to the stability, that is, the responsiveness in the non-DC region on the frequency region, which is common to control by broadcast transmission and independent distributed parallel processing. A dynamic compensator incorporating a delay model internally as well as the stability condition of passive characteristic priority enables to achieve a leap of responsiveness in control by many broadcast transmissions and independent distributed parallel processing. The present invention is surprisingly able to ensure high-speed responsiveness by adopting a controlled delay amount even if the delay on the broadcast system and the individual side is increased.

The target for which the total resources are limited is not limited to home power control. In air conditioning and lighting in the office, the application is very easy because the equipment, ie, the individual, is uniform. Of course, it is also targeted for the interconnection control with the power conditioner, which is composed of dwelling units, offices, and solar power generation facilities, which have the same power control, but also have solar cells and storage batteries mixed, and have different power consumption scales.

There are many situations in which resources are constrained. In data processing, although there is a certain amount of processing power on average, it often occurs when the amount of data generation exceeds the capacity at a certain point in time. It is necessary to continue the generation of high priority information, and to suppress the generation amount of low information. Although the problem has already been solved in the direct current region in Japanese Patent Application No. 2014-153348 and the international application WO 2015/115385, there is a delay, and ensuring stability in the non-direct current region over the frequency region The power control approach presented by the invention solves. FIG. 59 shows an example of data processing via a serial bus which enables information exchange between devices, which can be found in satellites, aircraft, automobiles and the like. In the recording apparatus or the communication apparatus, the resource of recording or transmission capacity, that is, processing capacity is limited, and it is shown a case where control of independent distribution of data generation amount is requested in parallel to devices which are individual groups. For example, in the communication apparatus, packets issued to the communication apparatus are sequentially temporarily stored and processed. When the margin for temporary storage decreases, broadcasts transmit a deviation or margin of processing capacity to the device group. However, since this transmission goes through a process of accumulating sequentially, it causes a delay of information transmission like a capacitor in the power system. That is, the point in time at which the communication device issues a processing power deviation is later than the point in time at which the device generates a packet. In addition, since the packets generated on the device side are sequentially transmitted, the transmission time is dispersed, and the delay amount is uncertain. Therefore, in a negative manner, a uniform reduction in packet generation amount is required to cope with the reduction in system sensitivity for the entire device. However, if the amount of multicast delay is managed in advance and delayed for transmission, and if the packets stored in the communication device are managed with the same time for each fixed time, a delay model is incorporated in the device side. By setting, high responsiveness can be maintained as shown by the effect of the invention. After the occurrence of the deviation, it is needless to say that it is too late, but if this method is adopted from the stage where the processing amount is within the processing capacity, by executing the preview control or PD control shown in the embodiment. It enables aggressive and fast resource allocation.
(Fig. 65 An example of a data processing apparatus which performs high-speed and stable processing by managing broadcast transmission and delay of individual side generated data)

The present invention is directed to a control that is effective when the delay in communication time, which relatively increases the amount of communication between the server and the client, limits the control speed, in the non-DC region on the frequency domain. Handles responsiveness. The obtained results, however, can be widely applied to independent distributed parallel control even in a system in which communication time is not limited. FIG. 60 is an example of a water supply system. This is, of course, also common to gas and refrigerant transportation for refrigerators and the like.
(In fact, if the communication time is high, but the population is composed of a large number of individuals, or if there are individuals that do not necessarily have a complete high-speed communication infrastructure, this fluid supply control rate In contrast, in many cases the communication time is not short, but in the industrial application described here, it can be commonly applied to a system with independent distributed parallel control, not in terms of communication time. Show)

Each supply individual, as well as the reservoir, which is the water source, has the inevitable characteristic of delayed response as well as the capacitor. For this reason, as for the flow rate sent by each supply individual at the destination, an event that is not measured by the main flowmeter in the measurement value occurs. This corresponds to the delay between the broadcast system and the individual. For this reason, in the independent distributed control, the supply suppression amount determined from the flow rate of the original is conservatively applied to each supply individual side by applying a small coefficient to suppress the water supply slowly. Immediate suppression causes instability, coupled with the measurement of the main flow. In each supply individual, stability and high speed can be achieved by setting a delay managed in advance and incorporating a delay model inside with other supply individual.
(Fig. 66 Application example for water supply)

Claims (14)

A function that coordinates the consumption of multiple resources with one individual that is exclusively responsible for the transmission of information to be shared, with the dynamic allocation of resources as a control goal, under conditions where total resources are regulated In a control system configured by independent distributed parallel processing individuals having
Based on the delay amount of each individual measured or estimated, the characteristic delay amount in the system is determined, and the condition for ensuring stability regarding the characteristic priority in the system is also determined.
A control system, characterized in that stability is ensured by multiplying the information that the broadcast individual broadcasts to the whole system, the system sensitivity, and adjusting the characteristic priority. In the previous term control system,
Information broadcast by the broadcast individual to the entire system is multiplied by the system sensitivity, and the phase can be manipulated, differentiated and integrated, or a combination thereof, or by a non-linear dynamic mechanism A control system characterized in that stability is ensured by adjusting the characteristic priority in a non-DC range by introducing a phase operation. In the previous term control system,
Information broadcast by the broadcast individual to the entire system is multiplied by the system sensitivity, and the phase can be manipulated, differentiated and integrated, or a combination thereof, or by a non-linear dynamic mechanism A control system characterized by satisfying the regulatory conditions of the extended resources by extending the resources to be controlled to the non-DC region by introducing phase operation. A function that coordinates the consumption of multiple resources with one individual that is exclusively responsible for the transmission of information to be shared, with the dynamic allocation of resources as a control goal, under conditions where total resources are regulated In a control system configured by independent distributed parallel processing individuals having
Based on the amount of delay of each individual measured or estimated, the only characteristic delay to be shared by the system is determined.
Define conditions for ensuring stability regarding characteristic priority in the system,
In order to ensure the characteristic delay amount relative to the broadcast information, each individual purposely adds an output delay mechanism independently and in parallel, and also a purposely appropriate dynamic delay model. A control system characterized by securing stability by incorporating. The control system according to any one of claims 1, 2, 3 and 4, wherein the resource to be regulated and allocated is power. The control system according to any one of claims 1, 2, 3 and 4, wherein the resource to be regulated and allocated is an amount of information per unit time. The control system according to any of claims 1, 2, 3 and 4, wherein the resources to be regulated and allocated are the flow rates of physically transported objects. A function that coordinates the consumption of multiple resources with one individual that is exclusively responsible for the transmission of information to be shared, with the dynamic allocation of resources as a control goal, under conditions where total resources are regulated In a control system configured by independent distributed parallel processing individuals having
Based on the delay amount of each individual measured or estimated, the characteristic delay amount in the system is determined, and the condition for ensuring stability regarding the characteristic priority in the system is also determined.
A method of control, characterized in that the information broadcasted by the broadcast individual to the entire system is multiplied by the system sensitivity and the characteristic priority is adjusted to ensure stability. In the previous term control system,
Information broadcast by the broadcast individual to the entire system is multiplied by the system sensitivity, and the phase can be manipulated, differentiated and integrated, or a combination thereof, or by a non-linear dynamic mechanism A control method characterized in that stability is ensured by adjusting the characteristic priority in a non-DC range by introducing phase operation. In the previous term control system,
Information broadcast by the broadcast individual to the entire system is multiplied by the system sensitivity, and the phase can be manipulated, differentiated and integrated, or a combination thereof, or by a non-linear dynamic mechanism A method of control, characterized in that the regulatory conditions of the extended resources are satisfied by extending the resources to be controlled to the non-DC range by introducing phase operation. A function that coordinates the consumption of multiple resources with one individual that is exclusively responsible for the transmission of information to be shared, with the dynamic allocation of resources as a control goal, under conditions where total resources are regulated In a control system configured by independent distributed parallel processing individuals having
Determine the amount of characteristic delay to be shared by the system based on the amount of delay of each individual measured or estimated
Define conditions for ensuring stability regarding characteristic priority in the system,
In order to ensure the characteristic delay amount relative to the broadcast information, each individual purposely adds an output delay mechanism independently and in parallel, and also a purposely appropriate dynamic delay model. A method of control characterized by ensuring stability by incorporating. The method of control according to claim 8, 9, 10, and 11, wherein the resource to be regulated and allocated is power.
The method of control according to claim 8, 9 or 10, wherein the resource to be regulated and allocated is the amount of information per unit time. The method of control according to claim 8, 9, 10, and 11, wherein the resources to be regulated and allocated are flow rates of physically transported objects.

Images