JP2009053938A - Equipment diagnosis system and equipment diagnosis method based on multiple models - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To generate diagnosis logic based on a highly reliable model wherein multiple models of diagnosis logic cooperate with each other for complementing. <P>SOLUTION: An equipment diagnosing system is provided with a plurality of monitoring devices 1-3 having databases 11, 21, 31 which store diagnosis logic based on a physical model, diagnosis logic based on a knowledge model, and diagnosis logic based on a statistical model, respectively. One of the monitoring devices, for example, the monitoring device 1 is provided with; a management processing part 17 which generates diagnosis logic based on a prescribed model and stores it in the database 11; and a communication means which transmits the generated diagnosis logic based on the prescribed model to the other monitoring devices 2, 3 which have the databases 21, 31 storing the multiple kinds of diagnosis logic based on the different models. The other monitoring devices 2, 3 are provided with management processing parts 27, 37 which convert the received multiple kinds of diagnosis logic based on the prescribed models to the different kinds of diagnosis logic based on the different prescribed models and store them in the databases 21, 31, respectively. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a facility diagnosis system based on a plurality of models and a facility for complementing the diagnosis logic based on the model while coordinating between the monitoring devices having the diagnosis logic based on different physical models, knowledge models, and statistical models. It relates to a diagnostic method.

  In general, as a facility diagnosis system using a computer system that diagnoses facilities based on measurement data observed in facilities (including plants, equipment, and systems), using diagnosis logic based on a desired model, There are many examples of equipment diagnosis. Here, the diagnosis includes preventive maintenance.

  Conventionally known models include physical models, knowledge models, statistical models, and the like.

  The diagnostic logic based on the physical model is suitable for predicting various events related to the behavior, performance, and failure of the equipment to be diagnosed by utilizing known physical information such as the physical information based on the equipment design information and physical phenomena. ing. As a typical system, there is a device remote diagnosis system that diagnoses a device failure using a simulator based on a physical model (Patent Document 1).

The use of diagnostic logic based on a physical model has the following advantages.
(1) The ability to use universally known physical laws such as mechanics, thermodynamics, energy and mass conservation laws.
(2) Ability to utilize foresight information about the target equipment such as equipment design information.
(3) It should be possible to make judgments using the laws of physics even if there is little measurement data observed at the target equipment.
(4) It is possible to predict even inexperienced events by executing a simulation based on a physical model.
(5) Events such as behavior, performance, and failure of the target equipment should be helpful to humans when making necessary decisions.
(6) Since the diagnosis logic based on the physical model is a physical law based on a physical phenomenon, it has an affinity with a designer or design information.

On the other hand, the disadvantage of diagnostic logic based on physical models is
(A) It is difficult to estimate parameters in the physical model.
(B) It is difficult to match the behavior of the physical model with the behavior of the actual equipment.
(C) The construction of the physical model may require technical knowledge and great effort.

  Next, the diagnosis logic based on the knowledge model is obtained by expressing the diagnosis logic in an easy-to-understand way for humans by accumulating knowledge, experience, etc. acquired from many years of operation and maintenance experience of the target equipment and other similar equipment. For example, if the type rule of an expert system which is a representative example of artificial intelligence, diagnosis logic expressed in failure mode effects analysis FMEA (Failure Mode Effects Analysis) format shown in FIG. 4, or failure tree analysis FTA (Failure Tree shown in FIG. 5) There is a method of expressing diagnostic logic expressed in (Analysis) format.

  Failure mode effect analysis FMEA is classified into large classification products, failure parts, failure modes, effects on failures, final results due to effects, failure occurrence probabilities, influence levels (levels) on large classification products, detectability, RPN (rotations per minute) ) Items are provided and are updated sequentially according to the operational state of the target equipment. The failure tree analysis FTA is used as a failure / accident analysis method. The failure tree analysis FTA is a method for determining the main cause by logically developing the occurrence conditions of a specific failure problem.

  As a specific application system, there is a failure diagnosis method by a computer system using diagnosis logic based on a knowledge model (Patent Document 2).

The use of diagnostic logic based on the knowledge model has the following advantages.
(1) By accumulating knowledge related to various events such as behavior, performance, and failure of the target equipment, it may be reusable for other similar equipment.
(2) Able to share knowledge and experience acquired by other similar equipment.
(3) It is a diagnostic logic expression closest to human judgment and has affinity with humans.

On the other hand, the fault of this diagnostic logic is
(A) Enormous amounts of data and necessary experience time are required to obtain knowledge about diagnostic logic based on the knowledge model.
(B) It is generally difficult to verify the validity and causal relationship of diagnostic logic. That is, when the characteristics of the target equipment change and the design method or manufacturing method is changed, it is difficult to make corrections unless the previous diagnostic logic can be used or similar data is collected.

  In addition, the diagnostic logic based on the statistical model is based on the sensors installed in the target equipment, the measurement data from the measuring device, the data related to operation / operation / control, and the environmental data related to equipment operation (weather, temperature, humidity, etc.) This is a method for predicting, judging, and diagnosing events such as behavior, performance, and failure of the target equipment by analyzing statistical data.

  As a specific application method, a preventive maintenance method and apparatus using diagnostic logic based on a knowledge model has been proposed (Patent Document 3).

The use of diagnostic logic based on this statistical model has the following advantages.
(1) The diagnostic logic can be generated automatically or semi-automatically based on the measurement data.
(2) By using the latest measurement data, it is possible to generate diagnostic logic that matches the behavior, performance, and failure of current equipment.
(3) Even if the characteristics of the equipment's behavior, performance, or failure change due to changes in the design of the target equipment or changes in the external environment, the measured data after the change is used, or causal relationships / correlations that affect the change. It is possible to generate an emergency statistical model including relevant external factor parameters.

(4) The diagnostic logic based on the statistical model has an affinity for the measurement data.

On the other hand, the diagnostic logic based on the statistical model has the following drawbacks.
(A) If missing data, sensor error data, or other erroneous data is mixed in the measurement data, it will hinder the generation of diagnostic logic.

(B) Since the diagnostic logic is generated only from the data measured by the sensor system, it is difficult to understand the essential meaning of the logic.

(C) Similarly, since the diagnostic logic is generated only from the data measured by the sensor system, the causal relationship between the mere correlation and the essential cause or result cannot be distinguished. For example, if there is a correlation between the data on the number of flying storks every year and the data on the birth rate of humans in the area, the strange hypothesis that if there are many storks, many children are born is estimated. Diagnosis logic is created that misses weather conditions, social phenomena, and other essential causal parameters.

(D) As the measurement data used as the basis for estimating the diagnosis logic based on the statistical model, the detection sensitivity of the diagnosis logic, the false detection rate, and the like vary depending on which measurement data is selected. Therefore, it is necessary to accumulate advanced engineering sense and trial and error experiments in selecting the measurement data as the basis.
Japanese Patent No. 3716318 Japanese Patent No. 3151093 Japanese Patent No. 3624546

  Therefore, the diagnostic logic based on the physical model, the diagnostic logic based on the knowledge model, and the diagnostic logic based on the statistical model have both advantages and disadvantages, and each diagnostic logic has a limit of application. Moreover, in order to overcome it, technical knowledge and great effort are required.

  In addition, in the maintenance of the target equipment, the design information, the information based on the physical laws, the knowledge model of the user and the operator, and the sensor data have been mutually organic, although a rational maintenance method is desired. This is an inefficient maintenance method because it is not used for any purpose, but only performs maintenance related to diagnosis and judgment according to individual diagnosis logic.

  The present invention has been made in view of the above circumstances, and is based on a plurality of models in which diagnosis logics based on a plurality of models complement each other in cooperation with each other, and generate diagnosis logic based on a model with higher confidence regarding diagnosis and judgment. An object of the present invention is to provide an equipment diagnosis system and an equipment diagnosis method therefor.

(1) In order to solve the above problems, in a facility diagnosis system having a plurality of monitoring devices for monitoring and controlling a target facility, diagnosis logic based on a physical model, diagnosis logic based on a knowledge model, and diagnosis logic based on a statistical model A plurality of monitoring devices each having a database for individually storing diagnostic logic based on at least two modems,
The one monitoring device generates first management processing means for generating diagnostic logic of a predetermined model by new, change or partial deletion and stores it in its own database; and the generated diagnostic logic of the predetermined model A communication means for transmitting the generated diagnostic logic of the predetermined model to the other monitoring device having a database that stores the diagnostic logic based on a different predetermined model after being stored in the database, The monitoring device is configured to include a second management processing means for converting the received diagnostic logic of the predetermined model into the diagnostic logic of the different predetermined model and storing it in its own database.

(2) The present invention provides a first monitoring device that monitors and controls the first target facility, a second monitoring device that monitors and controls the second target facility, and a third that monitors and controls the third target facility. And a monitoring center that receives measurement data observed by each monitoring device and monitors each target facility,
The monitoring center includes a diagnosis logic based on a physical model, a diagnosis logic based on a knowledge model, a diagnosis logic based on a statistical model, a database storing diagnosis logic based on at least two models, and a new, changed, or partially deleted When the diagnostic logic of any one model is generated and stored in a predetermined area of the database, the diagnostic logic of the other model is converted according to the generated diagnostic logic of the one model, A facility processing means for storing in a predetermined area and diagnosing the measurement data of each target facility observed by the monitoring device using a diagnosis logic based on a model associated with each monitoring device System.

(3) Moreover, this invention has each said database which each memorize | stores the diagnostic logic based on at least two modems among the diagnostic logic based on a physical model, the diagnostic logic based on a knowledge model, and the diagnostic logic based on a statistical model individually A plurality of monitoring devices and a monitoring center for collecting and monitoring at least the diagnostic results of the plurality of monitoring devices;
Each of the monitoring devices generates diagnostic logic based on a predetermined model by new, change, or partial deletion, stores the diagnostic logic in its own database, and then transmits the generated diagnostic logic based on the predetermined model to the monitoring center. The monitoring center includes means for transmitting diagnostic logic based on the predetermined model to the other monitoring device, and the other monitoring device receives the received diagnostic logic of the predetermined model by itself. It is an equipment diagnosis system provided with a management processing means for converting into the diagnosis logic of the different predetermined model possessed and storing it in its own database.

(4) Furthermore, the present invention is based on a physical model in a facility diagnosis method that monitors and controls each target facility and diagnoses the presence or absence of the corresponding target facility using a diagnosis logic based on a model owned by the present invention. Among the diagnosis logic, the diagnosis logic based on the knowledge model, and the diagnosis logic based on the statistical model, the plurality of monitoring devices each having a database each storing diagnosis logic based on at least two modems are provided,
A diagnostic logic generation step in which one of the monitoring devices generates diagnostic logic based on a predetermined model having a causal relationship with a cause / result in accordance with addition / deletion of a measurement sensor related to a target facility; and the one monitoring device includes: The diagnostic logic based on the generated predetermined model is transmitted to the other monitoring device having a different model of the diagnostic logic, and the causal relationship between the diagnostic logic of the predetermined model received by the other monitoring device. And converting and reflecting the diagnosis logic of the different predetermined model owned by itself.

  ADVANTAGE OF THE INVENTION According to this invention, the diagnostic logic based on multiple models which can complement and complement the diagnostic logic based on a plurality of models, and can generate diagnostic logic based on a model with higher certainty regarding diagnosis and judgment, and its diagnostic equipment method Can provide.

  DESCRIPTION OF EMBODIMENTS Embodiments of an equipment diagnosis system and an equipment diagnosis method based on a plurality of models according to the present invention will be described below with reference to the drawings.

  The subject of the present invention includes all plants, facilities, equipment, and systems such as facilities such as buildings and structures, various industrial plants and power generation plants, electrical equipment, computer systems, and communication systems. Hereinafter, these are collectively referred to as target equipment.

  In the following embodiments, a facility diagnosis system and a facility diagnosis method for performing performance diagnosis, abnormality diagnosis, and failure diagnosis that are commonly required for each target facility described above will be described.

(Embodiment 1)
FIG. 1 is a basic configuration diagram showing Embodiment 1 of an equipment diagnosis system based on a plurality of models according to the present invention.

  In the equipment diagnosis system based on the plurality of models, for example, a plurality of equipment monitoring devices (including servers, monitoring centers, local terminals, database devices, and other devices) 1 to 3 are installed. Databases 11, 21 and 31 are provided.

  The database 11 of the equipment monitoring apparatus 1 includes diagnostic logic and other necessary data based on a physical model generated based on known physical information such as design information of the target equipment 10 and physical laws based on physical phenomena as described above. Remembered. The database 21 of the equipment monitoring device 2 stores diagnostic logic and other necessary data based on the knowledge model generated in the form of the if the rule, for example, based on the operation and maintenance experience related to the target equipment 20 and the knowledge and experience acquired from the similar equipment. Is done. The database 31 of the facility monitoring device 3 stores sensors installed in the target facility 30, diagnostic logic based on a statistical model generated based on measurement data (actual data) measured by the measurement device, and other necessary data. . As each diagnosis logic, a conventionally well-known diagnosis logic is used in the initial setting stage.

  The facility monitoring devices 1 to 3 are connected to each other through linkage lines 4, 5, and 6. In other words, the facility monitoring device 1 and the facility monitoring device 2 exchange data with each other, and the facility monitoring device 2 and the facility monitoring device 3 exchange data with each other. The monitoring devices 1 are connected by a cooperation line 6 that exchanges data with each other. As these cooperation lines 4 to 6, a network such as the Internet, a WAN (Wide Area Network), and a LAN is used.

  The facility monitoring apparatus 1 generates a diagnostic logic based on a physical model and newly registers it in the database 11 while performing an input operation based on known physical information such as a physical law based on design information of the target facility 10 or a physical phenomenon. Or an input unit that stores in the database 11 after changing, deleting, or editing diagnostic logic based on a physical model in association with newly adding or deleting a measurement sensor at a required location of the target facility 10 12 and a display control unit that reads parameters such as a physical model and diagnostic logic currently registered in the database 11 according to various control instructions from the input unit 12, parameters of the model, and priority and displays them on the display unit 13. 14 and the diagnostic logic based on the corresponding model based on the diagnostic logic request from each equipment monitoring device 1 to 3 Alternatively, when the equipment monitoring devices 1 to 3 generate diagnostic logic based on a model with new, changed, or partially deleted, the communication unit 15 that automatically transmits diagnostic logic based on the generated model to another equipment monitoring device; An input / output interface 16 that collects necessary information from the target facility 10 and a management processing unit 17 are provided.

  In the equipment monitoring apparatuses 2 and 3, the input units 22 and 32, the display units 23 and 33, the display control units 24 and 34, the communication units 25 and 35, the input / output interfaces 26 and 36, and management are performed in the same manner. Processing units 27 and 37 are provided.

  The management processing unit 17 corrects or changes its own diagnostic logic in accordance with the diagnostic logic received from each of the other equipment monitoring devices 2 and 3, and the database 11 uses the change history of the corrected and changed diagnostic logic as log information. Are stored in a necessary area or an appropriate memory, managed, and stored as diagnostic logic based on a physical model in the database 11 and displayed on the display unit 13 via the display control unit 14 as necessary. The same processing is performed in the management processing units 27 and 37 of the other equipment monitoring apparatuses 2 and 3.

  The management processing unit 17 periodically transmits a diagnosis logic request signal to the other equipment monitoring devices 2 and 3 via the communication unit 15. For example, the other equipment monitoring devices 2 and 3 When there is a correction or change, a configuration may be adopted in which diagnostic logic based on a different model is received based on change notification information automatically issued from other equipment monitoring devices 2 and 3. The same processing is performed in the management processing units 27 and 37 of the other equipment monitoring apparatuses 2 and 3.

  In the equipment diagnosis system configured as described above, when a diagnosis logic based on a physical model is added, changed, or deleted in a specific equipment monitoring device, for example, 1, in response to a request from the other equipment monitoring devices 2 and 3, Alternatively, the information is automatically transmitted to the other equipment monitoring devices 2 and 3 through the communication unit 15 and the link lines 4 and 6.

  Each facility monitoring device 2, 3 receives diagnostic logic based on the transmitted physical model by the communication units 25, 35 and passes it to the management processing units 27, 37. Each management processing unit 27, 37 corrects or changes the diagnostic logic based on its own knowledge model and statistical model in accordance with the received diagnostic logic based on the physical model, and changes history of the diagnostic logic of this corrected and changed model. Is stored as log information in the required areas of the databases 21 and 31 and in an appropriate memory, and is stored in the database 11 as diagnostic logic based on the knowledge model and statistical model.

  In addition, the management processing units 27 and 37 will be described later when it is determined that the diagnosis logic of one of the management processors 27 and 37 is inconsistent with the received diagnosis logic, but in order to avoid the contradiction, correction and deletion of the diagnosis logic based on its own physical model are performed. I do. In addition, since only one diagnostic logic is input and set, diagnostic logics in other expressions are generated. Therefore, it is possible to complement each other so that there is no leakage between the diagnostic logics.

  The management processing unit 17 can also manage when an event change (event) occurs or by a timer. For example, at least the equipment monitoring device 3 periodically acquires measurement data (actual data). When the statistical model is updated each time it is acquired, the diagnostic logic is updated according to the updated model. The diagnostic logic based on the statistical model is automatically transmitted to the other equipment monitoring apparatuses 1 and 2. The management processing units 17 and 27 of the other equipment monitoring apparatuses 1 and 2 correct the diagnostic logic based on their own physical model and statistical model in accordance with the updated diagnostic logic.

  In addition, in the equipment monitoring device 2, when maintenance personnel inspect the target equipment on-site, newly obtained knowledge data is input and the knowledge model is updated. Then, the diagnostic logic is updated according to the updated model, but the diagnostic logic based on the updated knowledge model is automatically transmitted to the other equipment monitoring apparatuses 1 and 3. The management processing units 17 and 37 of the other equipment monitoring apparatuses 1 and 3 correct the diagnostic logic based on their own physical model and statistical model in accordance with the diagnostic logic based on the updated knowledge model.

  Further, when the design department changes the design parameters of the target equipment 10, when the CAD data is input by an input means such as the input unit 12 or other OCR, the management processing unit 17 updates the physical model and reflects it in the diagnosis logic. To do. The updated diagnostic logic is automatically transmitted to the other equipment monitoring devices 2 and 3. The management processing units 27 and 37 of the other equipment monitoring apparatuses 2 and 3 correct their diagnostic logic according to the updated diagnostic logic.

  Therefore, according to the embodiment as described above, the diagnosis logic based on the physical model, the diagnosis logic based on the knowledge model, the diagnosis logic based on the statistical model cooperate with each other, and the own diagnosis logic according to the other diagnosis logic By adopting a configuration that automatically converts, the following various effects can be obtained.

(1) When the diagnostic logic of one model is updated, it can be changed to a more reliable model diagnostic logic by automatically prompting the updating of the diagnostic logic of the other model according to the updated diagnostic logic. .

(2) When one diagnostic logic is updated, the diagnostic logic based on the latest model can always be secured by automatically prompting the updating of the diagnostic logic of another model according to the updated diagnostic logic.

(3) When a contradiction occurs between the diagnostic logics of the models, a part or all of the contradicting diagnostic logics can be eliminated or corrected.

(4) It is possible to eliminate redundant diagnostic logic that contradicts each diagnostic logic.
(5) The validity of each diagnostic logic can be checked.

  In the facility diagnosis system shown in FIG. 1, each of the facility monitoring apparatuses 1 to 3 is equipped with diagnosis logic based on different models. For example, as shown in FIG. , 20, 30, a physical model storage that stores at least diagnostic logic based on the physical model in the database 71 of the monitoring center 7 in a system to which a plurality of local terminals (local controllers, etc.) 8,. The system may include a unit 71a, a knowledge model storage unit 71b that stores diagnosis logic based on the knowledge model, and a statistical model storage unit 71c that stores diagnosis logic based on the statistical model.

In such an equipment diagnosis system, the monitoring center 7 has processing units corresponding to the management processing units 17, 27, and 37, and the processing unit receives control instruction information input from an input unit (not shown). Based on the diagnostic logic based on the generated physical model, when the diagnostic logic based on any one model, for example, a physical model, is generated by new, changed, or partly deleted, the knowledge model is different from the other model. And converted into diagnostic logic based on the statistical model, and stored in the knowledge model storage unit 71b and the statistical model storage unit 71c of the database 71. And the monitoring center 7
When the measurement data of each target facility 10, 20, 30 observed at the local terminals 8,... Is received, the target facilities 10, 20,. It diagnoses the presence or absence of 30 abnormalities.

  Moreover, as an equipment diagnosis system, as shown in FIG. 3, it is a system of the structure by which the equipment monitoring apparatuses 1-3 shown in FIG.

  In this equipment diagnosis system, the monitoring center 7 receives logic based on the model generated by the equipment monitoring apparatuses 1 to 3 and stores logic based on the model itself, while receiving the model received from other equipment monitoring apparatuses. It may be a system that transmits diagnostic logic, or converts the logic into a logic based on a model owned by another facility monitoring apparatus and transmits the logic to a corresponding facility monitoring apparatus.

  Hereinafter, the facility diagnosis system illustrated in FIG. 1 will be described as an example, but the system includes the facility diagnosis system illustrated in FIGS. 2 and 3.

  Furthermore, the facility diagnosis system shown in FIG. 1 can be realized as a facility diagnosis method based on a plurality of models.

The equipment diagnosis method based on a plurality of models is based on an equipment monitoring apparatus 1 having a database 11 for storing diagnosis logic based on a physical model, an equipment monitoring apparatus 2 having a database 21 for storing diagnosis logic based on a knowledge model, and a statistical model. When the equipment monitoring device 2 having the database 31 for storing diagnosis logic is provided,
One facility monitoring device, for example, 1 includes a diagnosis logic generation step for generating a diagnosis logic based on a predetermined model having a causal relationship regarding a cause / result as a measurement sensor for the target facility 10 is added / deleted, and this generation And transmitting the diagnostic logic based on the predetermined model to the other equipment monitoring devices 2 and 3 having the diagnostic model based on the statistical model and the knowledge model which are different models, and the other equipment monitoring device 2 3, when diagnostic logic based on a physical model is received, a variable relating to the causal relationship of the diagnostic logic of the model is taken out and converted into diagnostic logic based on a knowledge model or a statistical model. It can be done.

(Embodiment 2)
FIG. 6 is a diagnosis logic conversion diagram for explaining Embodiment 2 of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example of conversion from diagnostic logic based on a physical model to diagnostic logic based on a statistical model.

  It is assumed that the database 11 of the facility monitoring apparatus 1 defines a physical model 110 and a diagnostic logic 111 based on the model 110 as illustrated by, for example, new, changed, or partially deleted. As the structural information of the physical model 110, for example, a model for determining the door opening / closing time through a transmission system of a torque command-current control system-motor-belt-door (speed) -integration processing system for an elevator car door opening / closing control system. It has become.

  On the other hand, the if then rule is applied to the diagnosis logic 111 based on the physical model 110. That is, it is diagnostic logic for diagnosing a door drive system failure when there is a relationship of if estimated door opening / closing speed−measured door opening / closing speed> Va (speed threshold value) then.

  Therefore, the management processing unit 17 of the equipment monitoring apparatus 1 sends the request from the equipment monitoring apparatus 3 or automatically the structure information of the physical model 110, the information of the diagnosis logic 111 based on the model 110, and the diagnosis result information via the communication unit 15. To the equipment monitoring device 3.

  The management processing unit 37 of the equipment monitoring apparatus 3 takes in the structural information of the physical model 110 sent from the equipment monitoring apparatus 1, the information of the diagnostic logic 111 of the model 110 and the result information, and defines the statistical model defined in the database 31. 310 and structural information of diagnostic logic (not shown) based on the model.

Next, the operation of the present embodiment configured as described above will be described.
The management processing unit 17 of the equipment monitoring apparatus 1 relates to an elevator car door opening / closing control system, and performs door opening / closing control based on a torque command value. However, the disturbance (belt friction force) a and the disturbance (door deformation friction force) are controlled. ) Considering b, when the speed obtained by subtracting the actual measured door opening / closing speed from the predetermined estimated door opening / closing speed is smaller than the speed threshold Va, the actual door opening / closing speed is slow, so the door drive system is defective. Diagnose.

  Here, the management processing unit 17 sends the structure information of the physical model 110, the information of the diagnosis logic 111 based on the model 110, and the diagnosis result information via the communication unit 15 based on the request of the equipment monitoring device 3 or automatically. It transmits to the equipment monitoring device 3.

  When the management processing unit 37 of the equipment monitoring apparatus 3 receives the structural information of the physical model 110, the information of the diagnostic logic 111 based on the model 110, and the diagnostic result information via the communication unit 35, the conventional elevator that has already been defined. The statistical model 310 of the car door opening / closing control system is grasped, and the statistical model 311 of the car door opening / closing control system can be constructed even if the car vibration acceleration is unnecessary, compared to the statistical model 310 that receives the torque current and the car vibration acceleration. In addition, it is determined that only the door opening / closing speed lower limit value is sufficient as the output of the statistical model 310, and the input variable is corrected to a new statistical model 3111 limited to the torque current and the output variable to the door opening / closing speed lower limit value. In other words, variables of the conventional statistical model 310, for example, redundant variables such as car vibration acceleration and door opening / closing time are deleted, corrected to a simple statistical model 311 and registered in the database 31.

  In addition, the management processing unit 37 of the equipment monitoring apparatus 3 generates diagnostic logic 312 based on the corrected statistical model 311 instead of the existing diagnostic logic (not shown). That is, if the rule of if measurement door opening / closing speed <lower limit door opening / closing speed then, it is corrected to a diagnostic logic such as a door drive system failure and stored in the database 31 in the same manner. Thereafter, the presence or absence of door abnormality is diagnosed based on the rule of if measurement door opening / closing speed <lower limit door opening / closing speed then.

  Therefore, according to the embodiment described above, the structure is converted into the structure of the statistical model based on the physical model structure information, and the diagnostic logic 312 is generated based on the converted statistical model 311. It is possible to limit the diagnostic logic based on the minimum necessary logic, and it is possible to eliminate redundant diagnostic logic. The same effect as in the first embodiment can be obtained.

(Embodiment 3)
FIG. 7 is a diagnosis logic conversion diagram for explaining Embodiment 3 of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example in which diagnosis logic based on a physical model is converted to diagnosis logic based on a statistical model. More specifically, the diagnosis logic based on a statistical model is used in conjunction with the diagnosis logic based on a physical model. It is an example of diagnosing the presence or absence of abnormality in equipment.

  The database 11 of the equipment monitoring apparatus 1 includes a physical model 112 that combines a physical model 11a as shown and a target equipment (hereinafter referred to as an actual process if necessary) 10 by, for example, new or changed or partially deleted. It is assumed that the diagnostic logic 113 based on it is defined. The diagnostic logic 113 inputs a predetermined input signal to the physical model 112 and the actual process 10 and extracts a model error signal based on the difference between the measurement signal obtained from the actual process 10 and the output signal obtained from the physical model 112. Thus, the diagnostic logic 113 does not diagnose the presence / absence of the target facility 10 but can be said to be a logic for grasping the accuracy of the structure of the physical model (simulator) 112 with respect to the actual process 10.

  On the other hand, the database 31 of the equipment monitoring apparatus 3 defines a statistical model 313 for diagnosing the presence / absence of the target equipment and a diagnostic logic 314 based on the statistical model 313. That is, since the diagnosis logic 113 based on the physical model 112 stored in the database 31 of the equipment monitoring device 3 cannot diagnose the presence or absence of 10 abnormalities in the actual process, the equipment monitoring device 1 obtains the model error signal obtained from the diagnostic logic 113. Is transmitted to the equipment monitoring apparatus 3, and the presence or absence of abnormality in the actual process 10 is diagnosed by the statistical model 313 and the diagnostic logic 314 stored in the database 31 of the equipment monitoring apparatus 3. The equipment monitoring device 3 returns the diagnosis result to the equipment monitoring device 1.

Next, the operation of the embodiment configured as described above will be described.
The management processing unit 17 of the equipment monitoring apparatus 1 inputs a predetermined input signal at a predetermined cycle to the diagnosis logic 13 based on the physical model 112 including the actual process 10, and the behavior of the actual process 10 from the physical model (simulator) 112. Estimate the output variable for. On the other hand, a measured value is taken out from the actual process 11c.

  The diagnostic logic 113 based on the physical model (simulator) 112 outputs a difference signal between the calculated output value of the physical model (simulator) 112 and the output (measured value) of the actual process 10 as a model error signal.

  The management processing unit 17 of the equipment monitoring apparatus 1 acquires a model error signal that is a difference signal between the output calculation value of the physical model (simulator) 112 and the output (measurement value) of the actual process 10 from the diagnosis logic 113, and the actual process. Along with the output variables relating to the ten behaviors, it is transmitted to the equipment monitoring device 3.

  When the management processing unit 37 of the equipment monitoring apparatus 3 receives the output variable related to the behavior of the actual process 10 and the model error signal of the diagnostic logic 113, the management logic 37 uses the output variable related to the behavior of the actual process 10 to diagnose the logic 314 based on the statistical model 313. Is received, the received model error signal is input to the statistical model 313, and a discrimination signal indicating the presence or absence of an abnormality in the actual process 10 is acquired.

  The management processing unit 17 of the equipment monitoring apparatus 1 compares the discrimination signal obtained by the statistical model 313 with a predetermined threshold value by the diagnosis logic 314 based on the statistical model 313, and, for example, the absolute value of the discrimination signal is the threshold value. When the value is exceeded, an abnormality in the actual process 10 system is diagnosed.

  FIG. 8 is an example in which the diagnosis logic based on the physical model is further converted to the diagnosis logic based on the statistical model. Specifically, the target equipment is linked with the diagnosis logic based on the physical model and the diagnosis logic based on the statistical model. It is another example which diagnoses the presence or absence of abnormality.

  Examples of the physical model related to the facility monitoring apparatus 1 include a static model case and a dynamic model case. In the case of the static model, the physical model 1 is a combination of the model 114 and the actual process (target equipment) 10 using a software sensor, and measured values observed in the actual process 10 are input to the model 114 and the actual process 10. In this configuration, the state variable of the actual process 10 is estimated from the model 114 by feeding back the measurement value obtained from the actual process 10 to the model 114.

  On the other hand, in the case of the dynamic model, the model 115 using the state observer and the physical process 10 are combined, and the measured value observed in the real process 10 is input to the model 115 and the real process 10. In this configuration, the state variable of the actual process 10 is estimated from the model 115 by feeding back the measurement value obtained from the actual process 10 to the model 115.

  In the equipment monitoring apparatus 1, models 114 and 115 for estimating state variables of an unknown actual process 10 that cannot be measured in any case are stored in the database 11.

  The database 31 of the equipment monitoring apparatus 3 defines a diagnosis logic based on a statistical model for diagnosing the presence or absence of an abnormality in the actual process 10. That is, since the physical models 114 and 115 stored in the database 11 of the equipment monitoring device 1 cannot diagnose the presence or absence of the actual process 10, the state variable value obtained by the model 114 or 115 is transmitted to the equipment monitoring device 3. .

  The equipment monitoring apparatus 3 receives the state variable value as an input, acquires a determination signal indicating the presence or absence of an abnormality in the actual process 10 using the statistical model 315, and inputs the acquired determination signal to the diagnostic logic 316, It is the structure which diagnoses the presence or absence of 10 abnormalities.

  Next, the operation of the embodiment configured as described above will be described.

  The management processing unit 17 of the equipment monitoring apparatus 1 takes out the measured value from the actual process 10 and inputs it to the actual process 10 and the model 114 or 115, and estimates the state variable value of the actual process 11c from the model 114 or model 115. . At this stage, it cannot be diagnosed whether the actual process is abnormal.

  Therefore, when the management processing unit 17 of the facility monitoring apparatus 1 acquires the state variable of the real process 10 using the model 114 or the model 115, the management processor 17 transmits the state variable value of the actual process 10 to the facility monitoring apparatus 3.

  When the management processing unit 37 of the equipment monitoring apparatus 3 receives the state variable value acquired by the model 114 or 115, the state estimation value is input to the statistical model 315, and whether there is an abnormality in the actual process 10 from the statistical model 315. A discrimination signal representing is obtained.

  The management processing unit 17 of the equipment monitoring apparatus 1 compares the discrimination signal obtained by the statistical model 315 with a predetermined threshold value by the diagnosis logic 316 based on the statistical model 315, and for example, the absolute value of the discrimination signal is the threshold value. When the value is exceeded, it is diagnosed that the actual process 10 system is abnormal.

  Therefore, according to the embodiment as described above, in the equipment monitoring apparatus 1, the output variable (see FIG. 7) or the software sensor, the state observation regarding the behavior of the actual process 10 estimated using the simulator based on the physical model 112. When the internal state variable of the actual process 10 is acquired from the models 114 and 115 using a device and transmitted to the equipment monitoring device 3, the management processing unit 37 of the equipment monitoring device 3 uses the received output variable or internal state variable as a statistical model. 313, 315 and the statistical logic based diagnosis logic 314, 316 are taken out, and a discrimination signal capable of diagnosing the presence / absence of the actual process 10 is taken out, and the presence / absence of the abnormality in the actual process 10 is diagnosed. Diagnose using advanced statistical model 313, 315 and its diagnostic logic 314, 316b Bets can be, it is possible to obtain the same effect as in the first embodiment.

(Embodiment 4)
FIG. 9 is a diagnosis logic conversion diagram for explaining Embodiment 4 of an equipment diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example in which a diagnosis logic based on a physical model is converted to a diagnosis logic based on a statistical model. More specifically, in cooperation with a diagnosis logic based on a physical model, a sensor is used by a diagnosis logic based on a statistical model. This is an example of diagnosing the presence or absence of a system abnormality.

  The equipment monitoring apparatus 1 includes two physical models 116 and 117, an addition element 118 connected to the output lines of the physical model 116 and the physical model 117, and output lines of the physical models 116 and 117 and the addition element 118. A sensor diagnostic logic 122 based on a physical model composed of the connected measurement sensors 119 to 121 is generated and stored in the database 11. The two physical models 116 and the physical model 117 are constituted by physical laws that absolutely hold the state.

  As a result, when a certain input signal is input to both the physical models 116 and 117, mass generation is preserved by generating “Y1 = 2” from the physical model 116 and “Y2 = 3” from the physical model 117. Based on the above rule, “2” + “3” = “5” can be extracted from the addition element 118.

  The measurement sensors 119, 120, and 121 measure the outputs of the physical models 116, 117 and the summing element 118, but these measurement sensors 119, 120, and 121 usually have a property of gradually appearing as errors including drift. ing. However, the facility monitoring apparatus 1 cannot detect the abnormality of the measurement sensors 119, 120, and 121.

  Therefore, the management processing unit 17 of the equipment monitoring apparatus 1 transmits to the equipment monitoring apparatus 3 data related to the relational expression between variables or the constraint condition according to the physical model law in the diagnosis logics 116 and 117.

The equipment monitoring apparatus 3 is based on a statistical model using the physical models 116, 117 and the output values Y1, Y2, Y of the addition element 118 and the measured values Y1 ^* , Y2 ^* , Y ^* of the respective measurement sensors 119, 120, 121. A diagnostic logic 317 is generated and stored in the database 31, and the diagnostic logic 317 diagnoses whether there is an abnormality in the sensor system.

  Here, the diagnosis logic 317 based on the statistical model is the sensor installed in the actual process 10, the measurement data from the measurement device, the data related to operation / operation / control, and the environmental data related to facility operation as described above. This is because the diagnosis logic 317 based on the statistical model is suitable for (such as weather, temperature, and humidity). In other words, the diagnosis logic 317 based on the statistical model is used to predict, determine, and diagnose events such as behavior, performance, and failure of the actual process 10 by data analysis.

  Next, the operation of the present embodiment configured as described above will be described.

The management processing unit 17 of the equipment monitoring apparatus 1 reads the diagnostic logic incorporating the physical models 116 and 117 stored in the database 11, inputs a predetermined signal (for example, flow rate), the physical models 116 and 117, and the addition element The true flow rate values Y1, Y2, and Y are taken out from 118, and the measurement data Y1 ^* , Y2 ^* , and Y ^* are taken out from the respective measurement sensors 119, 120, and 121, and transmitted to the equipment monitoring device 3 through the communication unit 15, respectively. To do.

When the management processing unit 37 of the equipment monitoring device 3 receives Y1, Y2, Y, Y1 ^* , Y2 ^* , Y ^* from the equipment monitoring device 1, Y1 + Y2 which is a material (energy) balance constraint condition from Y1, Y2, Y = Y, and if each sensor measurement value Y1 ^* , Y2 ^* , Y ^* includes errors (variables) e1, e2, e, respectively, the constraint condition (Y1 ^* + e1) + (Y2 ^* + E2) = (Y ^* + e).

Therefore, the management processing unit 37 applies the least square method of min {e1 ² , e2 ² , e ² } and calculates the variables e1, e2, e by the diagnostic logic 317 based on the statistical model stored in the database 31. Then, the presence or absence of abnormality of the sensor is diagnosed from the relational expression of if min {e1 ² , e2 ² , e ² }> lower limit value (threshold value). That is, if there is a relationship of if min {e1 ² , e2 ² , e ² }> lower limit value (threshold value), it can be diagnosed that the sensor system is abnormal.

Further, when considering Y1 ^* ′ = (Y1 ^* + e1), Y2 ^* ′ = (Y2 ^* + e2), and Y ^* ′ = (Y ^* + e) with consideration of errors as other sensor measurement signals, other statistical models In the diagnosis logic based on the above, it is possible to perform a diagnosis based on the statistical model with high accuracy with respect to the sensor measurement signal in consideration of the error.

  Therefore, according to the embodiment as described above, the management processing unit 17 of the equipment monitoring device 1 extracts the relational data between the variables based on the laws of the physical model or the constraint condition data, and transmits the data to the equipment monitoring device 3. The management processing unit 37 of the equipment monitoring device 3 can correct a contradiction in data used for estimating the statistical model by making a diagnosis using a relational expression between the input and output variables of the statistical model or a constraint condition. This can detect a sensor error or make a diagnosis based on a statistical model with high accuracy by using measurement data in consideration of the sensor error, and has the same effect as the first embodiment.

(Embodiment 5)
FIG. 10 is a diagnosis logic conversion diagram for explaining the fifth embodiment of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example of conversion from diagnostic logic based on a physical model to diagnostic logic based on a knowledge model.

  The management processing unit 17 of the equipment monitoring apparatus 1 generates a physical model 123 relating to, for example, an elevator car door opening / closing control system based on data input from the input unit 12 and stores the physical model 123 in the database 11. Further, a cause-and-effect relationship model 124 is generated from the physical model 123 relating to the elevator car door opening / closing control system and stored in the database 11.

  Therefore, the generated physical model 123 and the causal relationship model 124 based on the physical model 123 are stored in the database 11 of the equipment monitoring apparatus 1.

  After generating the physical model 123 and the causal relation model 124, the management processing unit 17 of the equipment monitoring apparatus 1 reads out the physical model structure information and the causal relation model structure information, and based on a request from the equipment monitoring apparatus 2, or automatically To the equipment monitoring device 2.

  The database 21 of the equipment monitoring apparatus 2 stores the above-described FTA format diagnosis logic 210 shown in FIG. 5 and the FMEA format diagnosis logic model 211 shown in FIG. 4 related to the elevator car door opening / closing control system. The FTA format diagnosis logic 210 and the FMEA format diagnosis logic model 211 stored in the database 21 are accumulated based on the physical model structure information and the causal relationship model structure information before and after the present time transmitted from the equipment monitoring apparatus 1. It can be said that the FTA format diagnosis logic 210 and the FMEA format diagnosis logic model 211 of FIG.

  Therefore, when the management processing unit 27 of the equipment monitoring apparatus 2 receives the newly generated physical model structure information and causal relation model structure information from the equipment monitoring apparatus 1, the physical model is obtained from the physical model structure information and the causal relation model structure information. 113 and the causal relationship model 124 are analyzed and converted so as to be compatible with the conventional FTA format diagnosis logic 21a and FMEA format diagnosis logic models 210 and 211, and new FTA format diagnosis logic 210 and FMEA format diagnosis logic model 211 are generated. To do.

  The management processing unit 27 of the equipment monitoring apparatus 2 determines the priority order of the FTA format diagnosis logic 210 as shown by a dotted line on the basis of the physical model structure information and the causal relationship model structure information, and is, for example, a component from the drive system. For example, a door 212 is added and connected to a torque change.

  In addition, the management processing unit 27 of the equipment monitoring device 2 removes the component parts, if there are any components in the FTA format diagnosis logic 210 that are not related to the causal relationship model 124 relating to the elevator car door opening / closing control system. Process.

  Further, the management processing unit 27 of the facility monitoring apparatus 2 converts the FMEA format diagnosis logic model 211 based on the received structure information of the causal relationship model 124 while considering, for example, the degree of influence.

  Then, the management processing unit 17 of the equipment monitoring apparatus 1 diagnoses the elevator car door opening / closing control system by applying the physical model 123 and the causal relation model 124 based on the actual torque command value, and causes and results. Is analyzed. On the other hand, the management processing unit 27 of the equipment monitoring apparatus 2 also applies the new FTA format diagnosis logic 210 and FMEA format diagnosis logic model 211 based on the torque command value serving as the drive system, and diagnoses the cause and result. .

  Therefore, according to the embodiment as described above, based on the physical model 123 and the causal relationship model 124 generated by new or changed or partial deletion, the models of the diagnostic logic 210 and 211 based on the knowledge model are generated. The diagnosis logic models 210 and 211 in the database 21 that are physically inconsistent can be excluded.

In addition, the cause-cause relationship model 124 generated based on the physical phenomenon can be generated as a model of the diagnostic logic 210, 211 without omission. Further, it is possible to check the validity of the diagnostic logics 210 and 211 that are diagnostic logics of the existing knowledge model.
In addition, the same effects as those of the embodiment can be obtained.

(Embodiment 6)
FIG. 11 is a diagnosis logic conversion diagram for explaining the sixth embodiment of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example in which diagnosis logic based on a knowledge model is converted to diagnosis logic based on a statistical model.

  The management processing unit 27 of the equipment monitoring apparatus 2 is based on the diagnosis logic model 210 based on the knowledge model in the FTA expression format and the knowledge model in the FMEA expression format, based on an instruction from the input unit 22, by new or changed or partial deletion. A diagnostic logic model 211 is generated and stored in the database 21.

  The management processing unit 27 of the equipment monitoring device 2 generates a diagnostic logic model 210 based on the knowledge model in the FTA representation format and a diagnostic logic model 211 based on the knowledge model in the FMEA representation format, and then requests or automatically from the equipment monitoring device 3. The structure information of the diagnostic logics 210 and 211 is read out and transmitted to the equipment monitoring device 3.

  When the management processing unit 37 of the equipment monitoring apparatus 3 receives the structural information of the diagnostic logics 210 and 211, the diagnostic logic model 21c in the Bayesian Network format based on the structural information of the diagnostic logics 210 and 211 or already stored in the database 31. The basic structure of the system is generated, or in accordance with the structural information of the diagnostic logic models 21a and 21b, the learning logic model 318 already stored in the database 31 learns with the diagnostic logic model 318 and is monitored by the equipment monitoring device 2 or itself A transition probability (conditional probability density) between the components of the diagnostic logics 210 and 211 is estimated and learned from the failure history data of the equipment, and a diagnostic logic model 318 in a Bayesian Network format with higher accuracy is generated.

  This Bayesian Network is a type of neural network. When a causal relationship between structural information data of the diagnostic logics 210 and 211 is examined, when a certain component fails, another Bayerian Network is assigned to another component related to the certain component. Learn what happens as a transition probability.

  Therefore, the management processing unit 37 of the equipment monitoring apparatus 3 generates the Bayesian Network format diagnosis logic model 318 from the causal relationship of the structure information data of the diagnosis logics 210 and 211 and stores it in the database 21. In the figure, P1, P2,..., P11, P12,... Are transition establishments (weighting coefficients) of the connection between the component parts.

  Upon receiving the diagnosis logic result based on the knowledge model, the management processing unit 37 of the facility monitoring device 3 learns the result using the Bayesian Network format diagnosis logic model 318, and the facility monitoring device 2 again performs new diagnosis logic models 210 and 211. When the structural information data is received, it is learned by the Bayesian Network format diagnostic logic model 318. When a failure occurs in a component, the transition probability between connected components is changed, and the Bayesian Network format diagnostic logic is more accurate. A model 318 is generated.

  Upon receiving the measurement data from the target facility, the facility monitoring device 3 diagnoses the cause of failure and the failure result of the target facility according to the learned Bayesian Network format diagnosis logic model 318.

  Therefore, according to the embodiment as described above, the statistical model diagnosis logic 318 adapted to the failure history data is automatically used while considering the cause-effect relationship from the structure information of the diagnosis logic 210, 211 based on the knowledge model. Can be generated. Further, the same effects as those of the first embodiment described above are achieved.

(Embodiment 7)
FIG. 12 is a diagnosis logic conversion diagram for explaining Embodiment 7 of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example of conversion from diagnostic logic based on a knowledge model to diagnostic logic based on a physical model.

  In the management processing unit 27 of the equipment monitoring apparatus 2, the operator inputs necessary data from the input unit 22 based on past knowledge and experience, generates diagnostic logic 210 based on the knowledge model in the FTA expression format, and stores it in the database 21. Remember.

  The management processing unit 27 of the equipment monitoring apparatus 2 generates a diagnostic logic model 210 based on the knowledge model in the FTA expression format, and then reads out the request from the equipment monitoring apparatus 1 or the structural information of the diagnostic logic model 210 automatically. Transmit to the monitoring device 1.

  When the facility monitoring apparatus 1 receives the structure information of the diagnostic logic 210 from the facility monitoring apparatus 2, the facility monitoring apparatus 1 displays the structure information on the display unit 13.

  The database 11 of the equipment monitoring apparatus 1 or an appropriate memory stores a number of components related to the elevator car door opening / closing control system in advance.

  The management processing unit 17 of the equipment monitoring apparatus 1 automatically estimates the result (door opening / closing time) from the structural information of the diagnosis logic 210 based on the cause (drive system) and the drive system failure, and displays the result on the display unit 13.

  In this state, the operator on the equipment monitoring apparatus 1 side is based on the cause (driving system and torque change) and the result (door opening / closing time) and refers to the structural information displayed on the display unit 13 while referring to the elevator car. The torque command value from the drive system, which is the cause of the door opening / closing control system, is used as an input value, and it is selected from the components prepared in advance, and the current control system (parts)-motor (parts)-belt (parts) The physical model 123 is generated and stored in the database 11 by sequentially combining doors (parts) -integration processing elements (parts) and connecting the results (door opening / closing time). That is, when the management processing unit 17 of the equipment monitoring apparatus 1 receives the structural information of the diagnostic logic 210 from the equipment monitoring apparatus 2, a human system is interposed and the physical model 123 is generated semi-automatically.

  Therefore, according to the embodiment as described above, the physical model 123 used for the physical model diagnosis logic is appropriately and efficiently generated while considering the structure information of the cause and the causal relationship between the cause based on the knowledge model. be able to.

  In addition, a physical model 123 having a meaningful structure can be generated from a viewpoint based on experience and knowledge while taking various physical phenomena into consideration. Furthermore, unnecessary physical components of the physical model 123 can be eliminated and a simple physical model 123 can be generated.

(Embodiment 8)
FIG. 13 is a diagnosis logic conversion diagram for explaining an eighth embodiment of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example of conversion from diagnostic logic based on a statistical model to diagnostic logic based on a knowledge model.

  The management processing unit 37 of the equipment monitoring apparatus 3 refers to all failure history data including past operation result data relating to the elevator car door opening / closing control system, for example, and sequentially for each if then rule using the if then rule format. The conditional probability is estimated, a statistical model of the conditional probability is generated, and stored in the database 31. In the statistical model of conditional probability, A is, for example, torque normal, B is, for example, door opening / closing time is normal, Pab, Pcd: conditional probability, C is, for example, torque abnormality, and D is, for example, door opening / closing time abnormality.

  As a result, if the torque is normal A, the conditional probability that the door opening / closing time is normal B is 0.8. Further, if the torque abnormality C is present, the conditional probability that the door opening / closing time is normal D is 0.2. Here, all failure history data including past operation performance data is referred to, an if then rule format considered from these data is applied, a conditional probability is calculated, and a statistical model of the conditional probability is generated.

  The management processing unit 37 of the facility monitoring device 3 generates a statistical model of conditional probability, and then transmits the generated statistical model of conditional probability to the facility monitoring device 2 via the communication unit 35.

  When the management processing unit 27 of the equipment monitoring device 2 receives the statistical model of the conditional probability from the equipment monitoring device 3, it analyzes the statistical model of the conditional probability, and expresses the diagnosis logic 212 by the decision tree that is a kind of knowledge model. Is stored in the database 21.

  The management processing unit 37 of the equipment monitoring apparatus 3 uses, for example, failure history data including operation performance data that is sequentially accumulated every predetermined period, to add, change, or partially delete the conditional probability formula, thereby creating a new When the statistical model of the conditional probability is generated, it is transmitted to the equipment monitoring device 2.

  When the management processing unit 27 of the equipment monitoring device 2 receives the new statistical model of the conditional probability from the equipment monitoring device 3, it generates a new decision tree diagnostic logic 212 according to the statistical model of the conditional probability and stores it in the database 21. Remember.

  Therefore, according to the embodiment as described above, quantitative information and parameters (conditional probabilities) of causal relationships between causes and results based on statistical model diagnosis logic generated from failure history data including operation performance data. In consideration of the above, it is possible to generate a diagnosis logic of a knowledge model including a decision tree that is easy for humans to understand.

  Moreover, it is possible to adapt and tune diagnosis logic based on the knowledge model using failure history data including the latest operation result data at all times, and it is possible to generate diagnosis logic based on a more accurate knowledge model. In addition, the same effects as those of the first embodiment can be obtained.

(Embodiment 9)
FIG. 14 is a diagnosis logic conversion diagram for explaining Embodiment 9 of the facility diagnosis system based on a plurality of models according to the present invention.

  This embodiment is an example of conversion from diagnostic logic based on a statistical model to diagnostic logic based on a physical model.

  The management processing unit 37 of the facility monitoring apparatus 3 uses the operation history data of the target facility as an input signal, inputs it to the actual process 30 and a pre-designed system identification model (simulator) 320, and outputs the signal extracted from the actual process 30. Feedback is input to the system identification model 320. Then, in the system identification model 320, the system identification method is executed using the input signal and the output signal of the actual process 30, and the system identification model 320, which is a physical model having a causal relationship regarding the actual process 30, is generated. For example, using the operation history data of the power plant system, a parameter (for example, power generation efficiency) related to the behavior of the actual system 30 is estimated from the system identification model 320 which is a physical model by a system identification method.

  The management processing unit 37 of the equipment monitoring device 3 uses the system identification method and estimates a model parameter (for example, power generation efficiency) from the system identification model 320 that is a physical model, and transmits it to the equipment monitoring device 1.

  In the equipment monitoring apparatus 1, a model parameter storage unit 124 is provided in the database 11 or an appropriate memory, and the model parameter storage unit 124 includes rational model parameters of an actual system (for example, a power generation system) from a thermodynamic viewpoint. Is remembered.

  Therefore, upon receiving the model parameter 1215 of the system identification model 320 from the equipment monitoring device 3, the management processing unit 17 of the equipment monitoring device 1 inputs the model parameter 125 to the diagnosis logic 126 based on the physical model. The diagnosis logic 126 compares the model parameter 125 sent from the equipment monitoring device 3 with the model parameter that should be in the corresponding system stored in the model parameter storage unit 124, and diagnoses, for example, the state of power generation efficiency in the actual system 30. To do. For example, assuming that the power generation efficiency 43% is obtained as the model parameter 125 in the diagnosis logic 126 based on the physical model, and if the power generation efficiency is 40% at the maximum as the model parameter that should be, the actual plant has the aspect of the power generation efficiency. Diagnose abnormalities.

  Further, the cause of abnormality of the actual system 30 can be estimated from the diagnosis logic 126 based on the physical model.

  In the model parameter storage unit 124, a model parameter when the actual system is normal and a model parameter when the system is abnormal may be set, or may be a model parameter having upper and lower limit values.

  Therefore, according to the embodiment as described above, a model parameter suitable for the actual behavior data of the actual system can be acquired by the system identification function based on the statistical model. Further, since the diagnosis logic 126 based on the physical model is always diagnosed using model parameters based on the behavior of the real system, the optimum diagnosis result of the normal real system can be extracted, and is the same as in the first embodiment. There is an effect.

(Other embodiments)
FIG. 15 is a block diagram showing another embodiment of an equipment diagnosis system based on a plurality of models according to the present invention.

  In this embodiment, a plurality of facility monitoring devices 1, 2, 3 are connected to the monitoring center 7 via a network. The monitoring center 7 is provided with a database 71. The database 71 includes a physical model storage unit 71a, a knowledge model storage unit 71b, and a statistical model storage unit 71c, and a physical model display unit 72a and a knowledge model. Display section 72b and statistical model display section 72c are provided.

  On the other hand, the facility monitoring apparatus 1 has a function of monitoring and controlling the target facility 10 and generating diagnostic logic based on the physical model and storing it in the database 11 as described above. In addition, the facility monitoring apparatuses 2 and 3 also perform monitoring control of the target facilities 20 and 30, and generate diagnostic logic based on the knowledge model and diagnostic logic based on the statistical model as described above, and corresponding databases 21 and 31. It has a function to memorize.

  The monitoring center 7 is based on a model request, or diagnostic logic based on a physical model sent from each equipment monitoring device 1 to 3 at the time of model generation or periodically, diagnostic logic based on a knowledge model, diagnostic logic based on a statistical model Are stored in the corresponding storage units 71a to 71c of the database 71, and a multi-window environment is provided.

  That is, the monitoring center 7 is equipped with an OS adopting a GUI (Graphical User Interface), reads diagnostic logics 73a, 73b and 73c based on each model from the database 71 based on a display instruction input, and displays physical model display units 72a, 72a, The knowledge model display unit 72b and the statistical model display unit 72c can be simultaneously displayed and monitored.

  Then, for example, when a door which is a component element of the diagnostic logic 73a based on the physical model displayed on the physical model display unit 72 is clicked, it is determined as a door component element, and the knowledge model displayed on the knowledge model display unit 72b is displayed. The associated door 74 of the diagnostic logic 73a based is searched and blinked. Similarly, the diagnostic logic 73c based on the statistical model is also displayed blinking on a part associated with, for example, a door, which is a component element of the diagnostic logic 73a based on the physical model, and the correlation between the diagnostic logics 73a to 73c is monitored. Presented to members.

  As a result, the corresponding parts of the diagnostic logics 73a to 73c of each model can be displayed in a blinking manner or by color-coded display, so that they can be displayed in a display format that is easy for humans to understand. Can deepen.

  In addition, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the invention.

The block diagram which shows one Embodiment of the equipment diagnostic system based on the multiple model which concerns on this invention. The block diagram which shows other embodiment of the equipment diagnostic system based on the multiple model which concerns on this invention. The block diagram which shows other embodiment of the equipment diagnostic system based on the multiple model which concerns on this invention. The figure which shows the diagnostic logic expressed in failure mode influence analysis FMEA (Failure Mode Effects Analysis) format. The figure which shows the diagnostic logic expressed in failure tree analysis FTA (Failure Tree Analysis) format. The block diagram explaining the example converted into the diagnostic logic based on a statistical model from the diagnostic logic based on a physical model. The block diagram explaining the example which diagnoses the presence or absence of abnormality of a target installation by the diagnostic logic based on a statistical model, cooperating with the diagnostic logic based on a physical model. The block diagram explaining the other example which diagnoses the presence or absence of abnormality of a target installation by the diagnostic logic based on a statistical model, cooperating with the diagnostic logic based on a physical model. The block diagram explaining the example which diagnoses the abnormality presence or absence of a sensor system by the diagnostic logic based on a statistical model, cooperating with the diagnostic logic based on a physical model. The block diagram explaining the example converted into the diagnostic logic based on a knowledge model from the diagnostic logic based on a physical model. The block diagram explaining the example converted into the diagnostic logic based on a statistical model from the diagnostic logic based on a knowledge model. The block diagram explaining the example converted into the diagnostic logic based on a physical model from the diagnostic logic based on a knowledge model. The block diagram explaining the example converted into the diagnostic logic based on a knowledge model from the diagnostic logic based on a statistical model. The block diagram explaining the example converted from the diagnostic logic based on a statistical model into the diagnostic logic based on a physical model. The block diagram explaining the example which displays the correlation part of the diagnostic logic based on multiple models.

Explanation of symbols

  1-3 ... equipment monitoring device, 7 ... monitoring center, 8 ... local terminal, 10, 20, 30 ... target equipment (actual process), 11, 21, 31 ... database, 12, 22, 32 ... input unit, 13, 23, 33 ... display unit, 15, 25, 35 ... communication unit, 17, 27, 37 ... management processing unit, 71 ... database, 71a, 71b, 71c ... storage unit for each model, 72a, 72b, 72c ... each model Display unit.

Claims (13)

In equipment diagnosis system equipped with multiple monitoring devices to monitor and control the target equipment,
A plurality of monitoring devices each having a database for individually storing diagnosis logic based on at least two modems among diagnosis logic based on a physical model, diagnosis logic based on a knowledge model, and diagnosis logic based on a statistical model;
The one monitoring device includes: a first management processing unit that generates diagnostic logic of a predetermined model by new, change, or partial deletion and stores the diagnostic logic of the predetermined model in a different predetermined model; Providing communication means for transmitting to the other monitoring device having a database for storing diagnostic logic based on
The other monitoring device is provided with second management processing means for converting the received diagnostic logic of the predetermined model into the diagnostic logic of the different predetermined model and storing it in its own database. Diagnostic system. In equipment diagnosis system equipped with multiple monitoring devices to monitor and control the target equipment,
A first monitoring device that monitors and controls the first target facility; a second monitoring device that monitors and controls the second target facility; a third monitoring device that monitors and controls the third target facility; A monitoring center that receives measurement data observed by the monitoring device and monitors each target facility;
The monitoring center includes a diagnosis logic based on a physical model, a diagnosis logic based on a knowledge model, a diagnosis logic based on a statistical model, a database storing diagnosis logic based on at least two models, and a new, changed, or partially deleted When diagnostic logic of any one model is generated and stored in a predetermined area of the database, it is converted into diagnostic logic of another different model in accordance with the generated diagnostic logic of the one model, Management processing means for storing in the area,
A facility diagnosis system for diagnosing measurement data of each target facility observed by the monitoring device using a diagnosis logic based on a model associated with each monitoring device. In equipment diagnosis system equipped with multiple monitoring devices to monitor and control the target equipment,
The plurality of monitoring devices each having a database that individually stores at least two modem-based diagnosis logics among a diagnosis logic based on a physical model, a diagnosis logic based on a knowledge model, and a diagnosis logic based on a statistical model, A monitoring center for collecting and monitoring at least diagnostic results of the monitoring device;
Each of the monitoring devices is provided with means for transmitting diagnostic logic based on the generated predetermined model to the monitoring center after generating the diagnostic logic based on the predetermined model by new, changing or partial deletion and storing it in its own database. ,
The monitoring center includes means for transmitting diagnostic logic based on the predetermined model to the other monitoring device,
The other monitoring device is provided with management processing means for converting the received diagnostic logic of the predetermined model into the diagnostic logic of the different predetermined model owned by itself and storing it in its own database. Diagnostic system. In the equipment diagnosis method for monitoring and controlling each target equipment and diagnosing the presence or absence of the corresponding target equipment using diagnostic logic based on the model owned by itself,
The plurality of monitoring devices each having a database for individually storing diagnosis logic based on at least two modems among diagnosis logic based on a physical model, diagnosis logic based on a knowledge model, and diagnosis logic based on a statistical model are provided,
One of the monitoring devices generates a diagnostic logic based on a predetermined model having a causal relationship regarding a cause / result with addition / deletion of a measurement sensor related to the target facility,
Transmitting diagnostic logic based on a predetermined model generated by the one of the monitoring devices to the other monitoring device that owns the diagnostic logic of a different model;
A facility diagnosis based on a plurality of models, comprising: converting the diagnosis logic of the different predetermined model owned by itself based on the causal relationship of the diagnosis logic of the predetermined model received by the other monitoring device Method. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database storing diagnostic logic based on a physical model; and a second monitoring device having a database storing diagnostic logic based on a statistical model;
The first monitoring device includes a first management processing unit that generates diagnostic logic based on a physical model by new, change, or partial deletion, and the generated physical model structure information and the diagnostic logic information on the link line. And a communication means for transmitting to the second monitoring device through
The second monitoring device converts to the statistical model structure based on the received physical model structure information and the diagnostic logic information, generates diagnostic logic for the statistical model, and stores it in its own database. An equipment diagnosis system characterized by comprising a management processing means. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database for storing diagnostic logic based on a physical model generated by new, modification or partial deletion; and a second monitoring device having a database for storing diagnostic logic based on a statistical model. ,
The first monitoring device uses a simulator based on a physical model stored in its own database, and first management processing means for estimating an output variable or internal state variable related to the behavior of the first target facility; A communication means for transmitting the estimated output variable or internal state variable related to the behavior of the first target equipment to the second monitoring device through the link line;
The second monitoring device includes second management processing means for generating diagnostic logic based on the statistical model using the output variable or internal state variable received via the communication means and storing the diagnostic logic in its own database. An equipment diagnosis system characterized by being provided. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database for storing diagnostic logic based on a plurality of physical models generated by new, changing or partial deletion; and a second monitoring device having a database for storing diagnostic logic based on a statistical model With
The first monitoring device uses a plurality of physical models for an input signal, and obtains constraint condition data and variables based on the physical model law, and the management processing means A communication means for transmitting constraint data and variables,
The second monitoring device uses the constraint condition data and variables received via the communication means, generates diagnostic logic based on a statistical model that finds inconsistencies in the variables, and stores it in its own database. An equipment diagnosis system characterized in that a processing means is provided. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database for storing diagnostic logic based on a physical model generated by new, modification or partial deletion; and a second monitoring device having a database for storing diagnostic logic based on a knowledge model. ,
The first monitoring device includes a first management processing unit that generates a causal relationship model of a cause and a result based on the physical model from the physical model, and a cause and a result based on the physical model generated by the management processing unit. Communication means for transmitting the causal relationship model of the second to the second monitoring device,
The second monitoring device is a diagnostic logic based on a knowledge model having a priority expressed in an FTA format or FMEA format based on a cause-and-effect causal relationship model based on a physical model received via the communication means. An equipment diagnosis system characterized by comprising second management processing means for converting the data into a self-database and storing it in its own database. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database for storing diagnostic logic based on a knowledge model, and a second monitoring device having a database for storing diagnostic logic based on a statistical model;
The first monitoring device is generated by a first management processing unit that generates diagnostic logic based on a knowledge model expressed in the FTA format and FMEA format by new, change, or partial deletion, and the management processing unit. Communication means for transmitting a causal relationship model of diagnostic logic based on the acquired knowledge model to the second monitoring device,
The second monitoring device estimates a conditional random variable from measurement data observed at the target facility based on a causal relationship model of diagnostic logic based on a knowledge model received via the communication means, and is in a Bayesian Network format. A facility diagnosis system comprising a second management processing means for generating a diagnosis logic model based on a statistical model and storing it in its own database. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database for storing diagnostic logic based on a knowledge model; and a second monitoring device having a database for storing diagnostic logic based on a physical model;
The first monitoring device includes a first management processing unit that generates diagnostic logic based on the knowledge model expressed in the FTA format by new, change, or partial deletion, and stores the diagnostic logic in its own database, and the management processing unit Communication means for transmitting the structural information of the diagnostic logic based on the knowledge model expressed in the FTA format generated in step 2 to the second monitoring device,
The second monitoring device generates a physical model associated with the causal relationship between components based on the structural information of the diagnostic logic based on the knowledge model expressed in the FTA format received via the communication unit, and A facility diagnosis system, characterized in that second management processing means for storing in the database is provided. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database storing diagnostic logic based on a statistical model; and a second monitoring device having a database storing diagnostic logic based on a knowledge model;
The first monitoring device includes a first management processing unit that generates a statistical model diagnosis logic based on a conditional probability of the statistical model if the rule type using a measurement data group in which the correlation is confirmed; Communication means for transmitting statistical model diagnosis logic structure information related to the conditional probability generated by the processing means to the second monitoring device;
The second monitoring device generates diagnostic logic based on an if the rule rule knowledge model represented by a decision tree method from the statistical model diagnostic logic structure information related to the conditional probability received via the communication means, A facility diagnosis system, characterized in that second management processing means for storing in the database is provided. In the equipment diagnosis system according to claim 1 or claim 3,
A first monitoring device having a database for storing diagnostic logic based on a statistical model; and a second monitoring device having a database for storing diagnostic logic based on a physical model;
The first monitoring device includes a first management processing unit that estimates a parameter in a statistical model by a system identification method using a measurement data group whose correlation has been confirmed, and a statistic estimated by the management processing unit. Communication means for transmitting parameters in the model to the second monitoring device;
The second monitoring apparatus performs diagnosis based on a physical model that determines an abnormality from a long-term or short-term change tendency of the target facility based on a parameter in a statistical model received via the communication unit and a predetermined reference parameter. A facility diagnosis system comprising a second management processing means for generating logic and storing it in its own database. In the equipment diagnosis system according to claim 2 or claim 3,
The monitoring center is configured to simultaneously display diagnostic logic based on at least two models among diagnostic logic based on a physical model, diagnostic logic based on a knowledge model, and diagnostic logic based on a statistical model, and displayed by the monitoring means A facility diagnosis system comprising: means for designating a part of diagnosis logic based on one model to blink a designated related part of another model displayed by the monitor means.

Images