DE10013541A1 - Process adaptive optimization system for industrial communication system, changes system parameters based on communication system characters measured according to data traffic - Google Patents

Abstract

A controller (1) determines characters of industrial communication system based on data traffic detected by a monitor to provide instructions for changing starting and other system configuration parameters, with reference to operator's requirements.

Description

Industrial communication systems connect i. a. a variety of computers, Controls, actuators and sensors for the automated operation of technological Processes. The sensors and actuators distributed in the process are becoming increasingly distributed equipped with microprocessors. As a result of this decentralized data processing results there is an increased need for communication.

Technological processes generally place increased demands on industrial communication systems with regard to their requirements

• - Safety,
• - reliability,
• - fault tolerance,
• - real-time capability,
• - immunity
• - functional and topological expandability and
• - economical use.

This results in serious differences between the industrial ones Communication systems and communication systems used in telecommunications area.

Selection and construction of industrial communication systems in a special technological process take place i. a. such that the fulfillment of the alleged the requirements described above can be implemented as far as possible seems.

The decisive parameters for the operation of the communication systems are at commissioning is permanently set. Parameter changes during the runtime are at not provided in most industrial communication systems. That is, with yourself optimal system behavior can no longer meet changing requirements be guaranteed. Due to changing requirement profiles during operation, e.g. B. As a result of impending breakdowns in the systems, they are used Communication systems are increasingly pushed to the limits of their capabilities. In In extreme cases, these can collapse completely and possible accidents can no longer occur counteract.  

This static worst case parameterization makes it industrial Communication system oversized during runtime. Rising unexpectedly Requirements mean that the communication system is undersized.

On parameter changes in industrial communication systems during operation is currently due to the underlying complexity and competing goals omitted for parameter selection in almost all applications.

There is an important, singular one in the field of industrial communication systems Process that is used in extreme cases to increase reliability. This procedure only relates to increasing the availability of a Communication system. In the event of an error, there is a redundant one communication technology resource switched during operation as required.

In the publication atp automation automation practice 38 ( 1996 ), pp. 41 to 47, the structure of a redundant communication system is described as part of a technological process. In the publication, only this singular measure is described on a single communication system. There is no general approach.

In the publication Popp, M .: "PROFIBUS DP" (Hüthig Buch Verlag, Heidelberg. 1998) as a singular measure redundant strategies for an individual Communication system mentioned. A general approach for all industrial Communication systems do not exist in either publication.

Singular measures to avoid data transmission bottlenecks exist in the Area of telecommunications and wide area networks, which are pure computer networks represent. These relate to completely different security requirement profiles, Real time capability, economy etc. and network designs (continental and intercontinental communication) and computing resources (use of some High-performance computers) than in industrial communication systems fault, configuration and performance management. From technical and For economic reasons, these management functions, as in telecommunications field common, not direct to industrial communication systems transfer.

A method is known from DE 195 02 230 C2, which results from the observation a technical system, in particular a computer network, the adaptation of a fuzzy  Controller. This procedure is limited to a single parameter, the Orbital period of a token, depending on two measured values, the ring latency and the Load. In the process, the optimal design of a fuzzy controller is exclusive described for computer networks.

The network management levels used in the EN 50 254 standard relate to layers 1 and 2 of the OSI model and to layer 7 . A dynamic adaptation of the network is not possible. Network management services also exist in the EN 50170 standard, but do not perform dynamic self-optimization.

Another method for parameter optimization is described in the publication Fiedler, K .: "Time behavior analysis of serial bus systems using the load method". This The method relates to an optimization of the parameter Token-Target-Rotation-Time on the basis of simulation analyzes, but not on parameter optimization during the operation of an industrial communication system.

In the publication Kiefer, J .: "Methodical partitioning and parameterization of Fieldbuses "a parameterization of fieldbus systems is carried out. This is limited to an optimal selection of parameters based on a simulation of the technological process and the expected behavior of the Communication system. The genetic algorithm proposed in the publication is used for process simulation, based on this a single parameter of the Communication system and this only when commissioning the Communication system, but not during operation.

In the publication FET '99 Magdeburg, p. 342 to p. 350, Almeida, L. et al. "A flexible time-triggered communication system based on the controller area network: Experimental results "describes a flexible, time-triggered master / slave bus access. A planning computer determines beforehand whether a slave is to be included in a cycle are calculated and by sending a specific master call the affected slaves informed about this permission to access the bus. this concerns only one specific parameter - the sending permission of the slaves. In the present In contrast, according to the invention, several parameters are changed in a process-adaptive manner.  

Layers 1 and 2 of the OSI layer model are standardized in the international standard ISO 11898. The user of the communication system is responsible for setting up and using services for initialization and dynamic adaptation. The CANopen standard is based on the ISO 11898 standard and names 3 conceivable parameters that can be set during runtime: the identifier, the pause time of messages and the operating status of the connected participants. However, there are no methods for the process-adaptive change of these parameters in the sense of the present control system according to the invention.

The previously described process of redundant design of an industrial communication system for the extreme case of a resource failure initiated the search for solutions that not only allow configuration measures in the extreme case of a resource failure, but also a multifunctional control and optimization of industrial communication systems for numerous other critical ones or provide less than optimal traffic situations. The task was to develop and optimize a new type of control system ( 1 ) for industrial communication systems ( 2 ) with a variety of new intervention options that go far beyond the above-mentioned procedure.

This control system ( 1 ) should automatically adapt industrial communication systems ( 2 ) to the operationally fluctuating requirement profile (see above), which means that the requirement profile can be taken into account to a greater extent. The communication system is therefore regarded as a process to be controlled in the method.

The dynamic self-optimization method according to the invention was adapted to the limited computing power of the communication participants ( 19 .... 24 ) connected to industrial communication systems ( 2 ) and was not developed for use in telecommunications computer networks. Due to the central control of the dynamic self-optimization, only a small part of the computing power available in the communication participants ( 19... 24 ) is required for the parameter change. Self-optimization can be achieved inexpensively by using the computing resources available in the communication participants ( 19 ... 24 ).

The dynamic self-optimization of industrial communication systems ( 2 ) combines various, singular optimization interventions in optimization strategies ( 42 ... 47 ). The selection is made in a functional unit (selector ( 40 )) which contains an evaluation of the optimization strategies ( 42 ... 47 ) and precedes the optimization strategies ( 42 ... 47 ).

The method for dynamic self-optimization does not require a model of the technological process for the calculation. Only a model of the industrial communication system ( 2 ) used, which is stored in the communication system-specific control system ( 1 ), the communication-technical characteristics determined in the bus monitor ( 10 ) and their storage in a database ( 11 ) are required.

The method according to the invention reduces the use of commissioning and maintenance personnel to a minimum and reduces the downtimes of the technological process ( 5 ), which are caused by new parameter settings on the industrial communication system ( 2 ), by the early detection of a changed requirement profile and the immediately following automatic ones Parameter adjustment.

The method according to the invention does not violate any existing norm or standard of industrial communication systems. The structure proposed according to the invention is of a universal nature and can be used for any industrial communication system ( 2 ) due to the modular exchange of the bus monitor ( 10 ), the selective controller ( 12 ), the central ( 13 ) and the decentralized control device ( 27 ).

Principle of the new process

According to the invention, the industrial communication system ( 2 ) is viewed from a control-technical point of view as a controlled system ( FIG. 1), which presents itself as a variant over time.

The communication-technical characteristics of the industrial communication system ( 2 ) are determined by the technological process ( 5 ), which uses the industrial communication system ( 2 ) for data exchange, by its actual requirements ( 6 ). The operator ( 7 ), who has basic knowledge of the technological process ( 5 ), can make the basic settings of the industrial communication system (so-called start parameters) by appropriate parameterization. This can either be done directly by specifying the corresponding start parameters of the industrial communication system during the configuration phase, or indirectly (as in Fig. 1 ( 8 )) by transferring the basic knowledge of the technological process ( 5 ) to a storage system ( A possible embodiment is a database ( 11 ) and hereinafter referred to as such) (operator requirements ( 8 )), by means of which the start parameters are determined.

The invention relates to the data traffic-dependent control of an industrial communication system ( 2 ) by means of a control system ( 1 ) in accordance with important optimization criteria. The control system ( 1 ) determines the communication technology-relevant characteristics of the industrial communication system ( 2 ) by observing the data traffic ( 3 ) and gives configuration commands ( 4 ) to change the parameters and thus the behavior of the industrial communication system ( 2 ).

The commissioning of the industrial communication system ( 2 ) is carried out via the operator requirements ( 8 ) and corresponding start parameters. A bus monitor ( 10 ) ( FIG. 2) monitors and analyzes the data traffic taking place between the communication participants ( 19 .... 24 ) via communication lines ( 25 ). Communication participants ( 19... 24 ) are computers, controls, sensors and actuators. On the basis of the observation of the data traffic ( 3 ) and the data traffic analysis in corresponding functional units in the bus monitor ( 10 ), the communication technology-relevant characteristics corresponding to the actual requirements ( 6 ) of the technological process ( 5 ) are obtained. This information is stored in a database ( 11 ) by means of the connection ( 14 ). For data traffic analysis, relevant information stored in the database ( 11 ) is transferred to the bus monitor ( 10 ) via the connection ( 15 ).

The most appropriate optimization strategy ( 42 ... 47 ) is activated in the selective controller ( 12 ) ( FIGS. 2 and 4) using an upstream selector ( 40 ). The activated optimization strategy (e.g. 47) determines the necessary parameter changes and transfers them to a central control device ( 13 ), which transmits the parameter changes using configuration commands ( 4 ) ( FIG. 2). The decentralized control devices ( 27 ) in the communication participants ( 19 ... 24 ) recognize the required parameter change and take the control action. The communication participants ( 19... 24 ) now have an optimized parameter setting with regard to the requirements ( 6 ) of the technological process ( 5 ).

Furthermore, configuration commands ( 4 ) in the industrial communication system ( 2 ) can be used to control coupling elements ( 26 ) such that the topological structure of the industrial communication system ( 2 ) can be varied. Variation options are network segmentation, aggregation, combination and redundant network switchover to possible further communication lines. Corresponding coupling elements ( 26 ) and communication lines ( 25 ) must be provided in the topological structure of the industrial communication system in accordance with the topological strategies.

Structure of the database

The structure of the database ( 11 ) is such that it has global validity for all industrial communication systems ( 2 ) and technological processes ( 5 ).

The tasks of the database ( 11 ) consist in

• a) Recording of operator requirements ( 8 ) for determining the start parameters (the latter in the selective controller)
• b) Inclusion of all parameters that are to be set for all possible and stored operating states on the communication system (connection 18 )
• c) Recording of the currently arriving, communication-technically relevant characteristics (connection 14 )
• d) Provision of the data for the selective controller ( 12 ) and bus monitor ( 10 ) (connection 15 and 16 )
• e) Providing information about the industrial communication system for the operator (operator observation ( 9 )) (not required for process-adaptive optimization)

The structure of the database ( 11 ) is divided into three main areas:

• 1. I the classification tables ( 28 ),
• 2. II of the static configuration table ( 33 ) and
• 3. III of the dynamic requirements table ( 34 ).

The classification tables ( 28 ) include the

• 1. I network structure table ( 29 ),
• 2. II traffic table ( 30 ),
• 3. III participant table ( 31 ) and
• 4. IV message table ( 32 ).

The classification tables ( 28 ) correspond to selection lists with which the description of the operator requirements ( 8 ) is facilitated.

In these, sizes (classification features) are listed that are used for the communication-technical identification of industrial communication systems ( 2 ) and technological processes ( 5 ).

With the help of the features contained in the network structure table ( 29 ), the communication networks can be classified into different network structure classes. These features include the network topology, the communication medium, the maximum extent and the maximum number of connectable communication participants.

The characteristics in the traffic table ( 30 ) can be used to determine which types of traffic loads, such as. B. Operation at runtime, accident, batch processes etc., are generally to be provided.

The different classification features of communication participants ( 19... 24 ) are listed in the participant table ( 31 ). The main features of this participant table ( 31 ) are the maximum number of message sources and sinks per communication participant ( 19 ... 24 ), the priority and the diagnostic and parameterization ability of the communication participants ( 19 ... 24 ).

Another classification table is the message table ( 32 ). It can be used to assign the message sources of the individual communication participants ( 19... 24 ) to certain message classes with the associated characteristics. The features are e.g. B. the user data length, the message priority, the type of transmission (cyclic or acyclic), maximum waiting time and variance of the cycle time.

For the initialization of the database ( 11 ), the operator ( 7 ) must enter the operator requirements ( 8 ) in the static configuration table ( 33 ) with the aid of the described classification tables ( 29 ... 32 ) using the indirect method mentioned above.

The following procedure for creating the static configuration table ( 33 ) is proposed:

• a) Identification of the real network segments with the characteristics from the network structure table ( 29 ) with an indication of how many different data traffic loads exist per network segment (connection 35 ),
• b) Identification of all data traffic loads in all network segments with the characteristics specified in the data traffic table ( 30 ) with the indication of how many communication participants ( 19... 24 ) are assigned to each individual data traffic load (connection 36 ),
• c) Identification of all communication participants ( 19 ... 24 ) of all data traffic loads and network segments with the characteristics of the participant table ( 31 ) with an indication of how many message sources and sinks each individual communication participant ( 19 ... 24 ) has (connection 37 ),
• d) identification of the message sources of all communication participants ( 19 ... 24 ) from all data traffic loads and network segments with the characteristics in the message table ( 32 ) (connection 38 ),
• e) Linking the message sources of each individual communication participant ( 19... 24 ) with the associated message sinks in the other communication participants ( 19 ... 24 ), whereby this is also possible across network segments. It is not possible to link communication participants ( 19 ... 24 ) who are assigned to one network segment but different data traffic loads within the network segment.

The procedure according to points c) to e) has analogies to the creation of the communication relationship lists of the industrial communication systems ( 2 ). However, the communication relationship lists lack a lot of information, especially about the network segments and data traffic loads as well as the type of communication participants (19 24). The previously known communication relationship lists have no possibilities for dynamic self-optimization. The method of dynamic self-optimization according to the invention, on the other hand, enables continuous, process-adaptive optimization by updating the dynamic requirements table ( 34 ) during operation of the industrial communication system ( 2 ).

The process-specific information in the static configuration table ( 33 ) and the links to the classification tables ( 29... 32 ) contained therein are copied into an initial dynamic request table ( 34 ) after all operator requirements ( 8 ) have been entered (link 39 ). The configuration is now complete.

In addition to the connection ( 39 ) to the static configuration table ( 33 ), the dynamic request table ( 34 ) has connections ( 14 , 15 , 16 and 18 ) from and to the bus monitor ( 10 ) and the selective controller ( 12 ) and a connection for eventual operator monitoring ( 9 ). In addition to the operator requirements ( 8 ) and the communication characteristics that correspond to the actual requirements ( 6 ) of the technological process ( 5 ), the dynamic requirements table ( 34 ) also contains all parameter settings of the industrial communication system ( 2 ).

The parameter settings included include both the parameters of the current operating state and the parameters of the other possible operating states. The change between two operating states is simplified in that the parameters of all operating states are stored in the database ( 11 ).

Parameter changes made to the industrial communication system ( 2 ) are stored in the database ( 11 ) via the connection ( 18 ).

Structure of the selective controller

The selective controller ( 12 ) has a modular structure ( FIG. 4), the individual modules being dependent on the industrial communication system ( 2 ) to be controlled. The modules of the selective controller ( 12 ) are, on the one hand, an upstream selector ( 40 ) and, on the other hand, several optimization strategies ( 42 ... 47 ), of which only one is always activated via a switchover ( 41 ). The individual optimization strategies ( 42 ... 47 ) contain individual or a combination of several optimization interventions.

The selector ( 40 ) has two tasks: the evaluation of the characteristics of the industrial communication system ( 2 ) arriving from the database ( 11 ) via the connection ( 16 ) and the subsequent activation of a subsequent optimization strategy ( 42 ... 47 ).

For example, the load L, the actual (t _Wreal ) and the allowed (t _Wgeg ) waiting time of messages N arrive in the selector ( 40 ) as characteristics of the industrial communication system ( 2 ). The load L is understood to mean the messages N sent in a specific time period t. These values are evaluated by the selector ( 40 ).

If there is a case of a high load L and the waiting times (t _Wreal <t _Wgeg ) of all messages N are _exceeded , z. B. the increase in the gross data rate as an optimization strategy (z. B. 42) received a high rating and thus activated. If, on the other hand, there is a low load L and only the exceeding of individual permitted waiting times can be recorded, various optimization strategies ( 42 ... 47 ) are available. These strategies can include optimization interventions for changing the bus access, changing the priority for the relevant messages N _i or a combination of these optimization interventions. The selection of the most appropriate optimization strategy ( 42 ... 47 ) is based on the evaluation in the selector ( 40 ).

The selector ( 40 ) therefore has knowledge of the behavior of the industrial communication system ( 2 ) as well as of the subsequent optimization strategies ( 42 ... 47 ).

A suitable combination of singular optimization interventions creates new optimization strategies ( 42 ... 47 ) that enable the industrial communication system ( 2 ) to be quickly adapted to changing requirements ( 6 ) of the technological process ( 5 ).

Each individual optimization strategy ( 42 ... 47 ) is structured in such a way that it does not contain any competing goals. The selection by the selector ( 40 ) of only a single optimization strategy ( 42 ... 47 ) prevents the optimization interventions that may serve competing goals from interfering with one another.

The structure and the concrete implementation of the optimization strategies ( 42 ... 47 ) in the selective controller ( 12 ) are to be selected on the basis of a control engineering analysis of the industrial communication system ( 2 ). Can be used for. B. the classic multivariable controllers, fuzzy controllers, controllers based on neural networks or Petri networks, etc. The use of elementary controllers, such as a two-point controller, is also possible.

The optimization strategy (e.g. 47) activated by the selector ( 40 ) via the switchover ( 41 ) transfers the calculated parameters via the connection ( 17 ) to the central control device ( 13 ). This sets the parameter information in configuration commands ( 4 ), which one to all communication participants ( 19 ... 24 ) or coupling elements ( 26 ) receive in the decentralized control device ( 27 ), evaluate and then change the corresponding parameters. The parameter change is stored in the database ( 11 ) via the connection ( 18 ).

Description of the optimization interventions

The optimization interventions described below can be contained individually or in a suitable combination depending on the technological process ( 2 ) in the optimization strategies ( 42 ... 47 ) in the selective controller ( 12 ) of the control system ( 1 ).

Change in gross data rate

A change in the gross data rate becomes necessary if the load L of the data volume exceeds a maximum value L _max or falls below a minimum value L _min . The maximum load L _max usually means that many participants have to wait for a transmission permission.

The gross data rate of industrial communication systems ( 2 ) is an essential parameter for adaptation. Various properties of the communication system can be influenced with this parameter. Such properties include the maximum possible length of the communication lines ( 25 ), the distance between sending and receiving communication participants ( 19 ... 24 ), the transmission time for a telegram and the resulting cycle time for a telegram cycle and the sensitivity to data transmission, e.g. B. against electromagnetic interference.

In previous industrial communication systems, the operator ( 7 ) predefines the gross data rate when the industrial communication system ( 2 ) is put into operation, although this should be controlled as an essential parameter in a process-adaptive manner.

So it can happen that messages due to a too low gross data rate have to wait too long or that the interference sensitivity of the telegrams due to a too high data rate is too high.

A certain minimum of the gross data rate is allowed in certain applications for ensuring a time-equidistant regulation not undercut as well as a Maximum due to the permitted cable length of the communication system be crossed, be exceeded, be passed. The gross data rate can also be optimally taken into account Noise sensitivity of the data transmission is only set to be process-adaptive, i.e. at runtime become.

Prioritization

Optimization interventions to change the priority of communication participants ( 19 ... 24 ) or message sources should be used if high-priority communication participants ( 19 ... 24 ) or messages to lower priorities have to wait for bus access.

The communication participants ( 19 ... 24 ) and the transmission messages can be assigned priorities for the transmission of data by means of various parameters, e.g. B. by:

• - the polling frequency of slaves with a master / slave bus access,  
• - changing the token target rotation time for token ring and token bus Systems,
• - The allocation of prioritized identifiers for event-oriented bus access and
• - the limitation of the load of individual participants.

Change of bus access

The optimization interventions to change the bus access can lead to a rapid adaptation of the industrial communication system ( 2 ) for the transition between different operating states in the technological process ( 5 ).

In the transition from a cyclic data exchange at runtime to an acyclic data exchange in the event of an accident, where high-priority warning and error messages and the resulting reaction commands have absolute priority over other data of the technological process ( 5 ), a change from master / slave to event-oriented bus access significantly reduce the waiting time t _{Wi of} the few high-priority messages N _i .

A further example of the use of an optimization intervention to change the bus access would be a high load L for the communication system, which leads to individual, low-priority messages N _j waiting for a significant time t _Wj in an event-oriented bus access or, in extreme cases, not being allowed to send. By switching to a master / slave bus access, these messages N _{j can} now be granted access. For the other messages N _i , this results in an increased waiting time t _Wi .

In this example, several optimization interventions compete with one another respective Optima cannot be fulfilled at the same time. The activated and set Optimization strategy (e.g. 47) fulfills the task of finding the most appropriate parameter setting to find.

Bus access cannot be changed in all industrial communication systems because the existing norms or standards prohibit this.

Fast adaptation to discontinuous changes in requirements

Another optimization intervention is the adaptation to different driving styles of the technological process ( 5 ). These can result from different operating states of the technological process ( 5 ), such as. B. Accident cases that require completely different parameter settings from the industrial communication system ( 2 ) than they are necessary for a cyclic data exchange during the runtime.

So far, these z. T. widely differing requirements for the industrial communication system ( 2 ) covered by a worst-case parameter setting. This is avoided by the dynamic self-optimization of the communication system proposed in the invention. Now optimal communication can be ensured during all operating phases.

For a quick switchover, it is necessary that the control system ( 1 ) sends a few configuration commands ( 4 ) to the decentralized control devices ( 27 ) of all communication participants ( 19 .... 24 ) via the central control device ( 13 ).

In order to keep the load on the industrial communication system ( 2 ) for changing the parameters as low as possible, the decentralized control devices ( 27 ) have lists. The lists contain information on how the communication subscriber ( 19... 24 ) has to behave during certain operating states. These lists were initialized by the control system ( 1 ) during the commissioning phase and updated during the entire operating time.

In the case of the invention, only one configuration command ( 4 ) then needs to be sent from the central actuating device ( 13 ) to all decentralized actuating devices ( 27 ). This configuration command ( 4 ) only contains a clear setting command for switching to the desired target list, which contains the parameters for the special, new operating state.

Adaptive participant selection

The cycle time can be considerably reduced by selecting the communication participants ( 19 ... 24 ) required for data exchange, because only process-relevant messages N are always sent by the participants.

With this optimization intervention, certain communication participants ( 19... 24 ) can be selectively separated from the data exchange within certain operating states.

Load limitation, to avoid load-related failures

Optimization interventions, which include the limitation of the load L, can avoid load-related failures of the industrial communication system ( 2 ). The load L is understood to mean the messages N sent in a specific time period t. The load of a subscriber L _i results from the number of his send messages N _i . The number of transmission messages N _i , in a certain period of time t is limited in the load limitation.

Another intervention is the change in the pause times t _{Pi of} individual news sources i. These message sources can only send a new message N _i after the pause time t _{Pi has} expired. However, this also affects the real-time capability of this message N _i . There are therefore competing optimization strategies for the pause time t _Pi of messages N _i , which must be evaluated by the selector ( 40 ) mentioned. The optimization strategy (increasing the waiting time or the real-time capability of a message) with the larger rating factor is activated by the selector ( 40 ).

Diagnosis of the participants

Another possibility to improve the performance is the constant monitoring of the communication participants ( 19 ... 24 ) by the selective controller ( 12 ). Based on the available data, the latter has the option of detecting early failures of the communication participants ( 19 ... 24 ) and reacting to them by changing the parameters.

This constant monitoring of the communication participants ( 19 ... 24 ) by a bus monitor ( 10 ) already exists in some cases. In many cases, however, the intervention of the maintenance personnel is still required to change the parameters on the industrial communication system ( 2 ). The combination of subscriber diagnosis with further adaptation interventions also opens up new possibilities for increasing the reliability of an industrial communication system ( 2 ).

Of course, the selective controller ( 12 ) cannot repair or replace a defective communication participant (e.g. 24), but it can adapt the remaining communication participants ( 19 ... 23 ) to this situation by changing parameters. Furthermore, the operator ( 7 ) can recognize the deterioration in the operating state of individual communication participants ( 19.... 24 ) through the statistical evaluation in the database ( 11 ).

The previous optimization interventions presented an improvement in the behavior of industrial communication systems on the basis of parameter changes within the communication participants ( 19 ... 24 ). Furthermore, the coupling elements ( 26 ) can also be changed in their communication-technical behavior by optimization interventions, on the one hand by the above-mentioned optimization interventions or combinations of them and / or on the other hand through new structure-changing optimization interventions. The latter are shown below.

Adaptation of the topology, network segmentation and redundancy switching

According to the invention, the process-dependent structure change of the industrial communication system ( 2 ) is carried out by one or more structuring elements, so-called coupling elements ( 26 ), which are changed depending on the process by commands from the control system ( 1 ) in such a way that the data traffic is in the most suitable form depending on the data traffic situation can be carried out.

Data traffic situations that are predestined to be mastered and optimized by a structural change in the communication system are:

• - general overload situations,
• - Need to increase the data rate over a value that is appropriate for the Overall expansion of the entire communication system is too high,
• - error situations

The construction of a coupling element ( 26 ) which can be switched as a function of data traffic and which resembles a bridge is shown by way of example in FIG. 5. In addition to the bridge, the coupling element ( 26 ) offers two intervention options for controlling its behavior depending on the traffic. In contrast, previously known bridge circuits do not offer such intervention options.

Known bridges that only have a fixed setting can Both accelerate and impair data traffic: a bridge to date In a network with a high network load, this can be done by its segmentation effect Reduce data traffic in the emerging segments considerably and thus for one ensure faster data throughput in the relieved segments. The same works against it Bridge in the same network slowing down to low traffic Traffic, which is a huge disadvantage.  

You have this disadvantage in previous communication systems even if you tried the i. a. only temporary traffic overload situations counter by having the data rate as high as possible and economically reasonable. However, this then works during the entire operation a communication system, even at times of low traffic.

Apart from the disadvantages of high data rates, e.g. B. by interference a high data rate can only be achieved by dividing the network into shorter network segments to reach. This leads to the compulsion that these network segments - with the disadvantage described - to couple through bridges.

According to the invention, the disadvantages described are eliminated by the control system ( 1 ) adapting the coupling element ( 26 ) to the traffic volume and / or the data rates, activating it in the case of higher traffic volume and / or higher data rates and deactivating it in the case of low traffic volume and / or low data rates.

Configuration commands ( 4 ) from the control system ( 1 ) are identified after passage through a left or right receiver ( 48 or 49 ) in a corresponding command receiver ( 50 or 51 ) for latching the coupling element ( 26 ) and trigger the activation of the coupling element ( 26 ) off (coupling element activation 52 or 53 ). The message telegrams N ₂₂ , N ₂₃ , N ₂₄ or N ₁₉ , N ₂₀ , N ₂₁ following the configuration commands ( 4 ), those for the communication participants ( 22 ... 24 or 19... 21 ) are stored in a register ( 54 or 55 ) and, after being processed by filter, memory, priority and actuator management ( 56 or 57 ), transferred to a queue ( 58 or 59 ). From here, the messages are forwarded via a transmitter ( 60 or 61 ) to the other side of the coupling element ( 26 ).

The optimization intervention on the coupling element ( 26 ), which can be used to change the structure of an industrial communication system ( 2 ) during its operation, is also intended to serve as an example of an intervention that is predestined for the combination with other optimization interventions to implement a more complex optimization strategy: In the control system ( 1 ) used for self-optimization of the industrial communication system ( 2 ), a decision is made according to a combination of weighted individual interventions, whether in the event of an overload situation

• the activation of coupling elements ( 26 ) without increasing the gross data rate,
• - The activation of coupling elements ( 26 ) with an increase in the gross data rate or
• - No activation of coupling elements ( 26 ) and only an increase in the gross data rate

he follows.

In addition, a switching element circuit according to the invention (shown by way of example in FIG. 5) offers the possibility of a further data traffic-dependent optimization intervention, the so-called "filter, memory, priority and control device management" ( 56 or 57 ), in that the messages that come from the left to the right (or vice versa) are transported through the coupling element, depending on the data traffic

• - selected and / or
• - can be rearranged in their priority.

Decentralized control devices ( 62 ) are also part of the functional units of management. Similar to the setting devices ( 27 ) in the communication participants ( 19 ... 24 ), they can ( 62 ) on the part of the control system ( 1 ) from one of the above-mentioned parameter-influencing optimization interventions or a weighted combination of these according to a specific optimization strategy for carrying out a corresponding setting of the Coupling element ( 26 ) are caused.

An option for the immediate return of an acknowledgment message in the direction the system from which a message was received is given.

The control system ( 1 ), which optimizes the industrial communication system ( 2 ), can be placed on the left as well as on the right of the coupling element ( 26 ): In any case, a command arriving at the coupling element ( 26 ) both causes the coupling element ( 26 ) to engage on the left ( 48 and 50 ) as well as on the right ( 49 and 51 ) receiver of the coupling element the switchover required there.

A further optimization intervention is provided by the following network segmentation depending on errors in the communication system, which enables: In the event of errors in individual modules, these modules are separated from the communication system, which has a suitable modular structure, by:
either the remaining modules remain in a temporarily sufficient mode without further interference from the interfering modules, each of them still capable of communicating
or the disruptive modules are replaced by equivalent modules in the "hotstandby" or "coldstandby" in the network.

Embodiment

The invention is carried out by means of the industrial communication system CAN ( 2 ), which offers a large number of parameter changes during operation.

In the tested embodiment, the industrial communication system ( 2 ) consists of 8 communication participants ( 19 ... 24 ) and a control system ( 1 ). The influencing of the industrial communication system CAN ( 2 ) with variable, actual requirements ( 6 ) of a technological process ( 5 ) in real communication participants ( 19 ... 24 ) is simulated by a changing load and different operating states of a model technological process ( 5 ).

Commercially available CAN controllers are used as communication participants ( 19... 24 ), which are connected to a programmable micro-controller. Each individual micro-controller contains a sub-program which contains the decentralized control device ( 27 ) for changing the communication parameters. In terms of structure, the communication participants ( 19... 24 ) used in the exemplary embodiment correspond to the participants in the typical industrial environment.

The control system ( 1 ) is located in a PC that contains a CAN PC plug-in card. The control system ( 1 ), which contains the bus monitor ( 10 ), the selective controller ( 12 ) and the central control device ( 13 ) as well as the connections ( 14 ... 18 ) shown in FIG. 2, is combined in one program. The database ( 11 ) is also on this PC. The operator (7) can then through the database (11) enter on this PC operators claims (8) or the state of the industrial communication system (2) to query (operator observation (9)).

The bus monitor ( 10 ) accesses the data traffic ( 3 ) received from the CAN PC plug-in card. The configuration commands ( 4 ) to be sent by the central control device ( 13 ) are also sent via the same CAN PC plug-in card.

Description of the implemented optimization interventions Change in gross data rate

The gross data rate is changed after the recognition of a changed load L or after requesting a different gross data rate from a communication subscriber ( 19 ... 24 ).

Mechanisms and interlocks are provided in the embodiment of the invention, that prevent data loss when the gross data rate changes.

Change the identifier

With CAN, the priority of a message source i is determined by means of the identifier Access to the communication medium specified. This identifier determines which other news sources take precedence over the respective news source. The news source with the highest priority always gets access as soon as that Communication medium is free.

The messages are automatically ranked by the identifier. The ranking is based on various criteria. These include the following criteria:

• - the type of information contained (high priority alarm messages or lower priority life reports),
• - the maximum allowed waiting time for a message,
• - the length of the message,
• - The frequency of sending a message and / or
• - The importance of the participant (e.g. alarm sensor).

The choice of the identifier for a message is therefore a multi-criteria decision. I. d. This priority will only be assigned during the commissioning of the CAN carried out. The method according to the invention enables the modification of Identifiers during operation.  

Change of bus access

There are two types of bus access, one of them deterministic and on the other hand probabilistic bus access. Both have and disadvantages that can be found in the corresponding specialist literature.

In the structure according to the invention, the two types of bus access are switched smoothly. The selective controller ( 12 ) accesses the following parameters:

• - Type of bus access (switching between master-slave and event-oriented)
• - contained news sources in one cycle
• - Identifier of the message
• - Minimum waiting time for a message when it is sent again

Through a graded variation of these parameters, the CAN can be strictly Deterministic bus access gradually converted to event-oriented access become.

Change the cycle time

The cycle time t _Z indicates how long a data exchange cycle lasts. Various types of intervention are used to change the cycle time. With the optimization interventions described above, the cycle time t _{Z can also be} dynamically adjusted in addition to the bus access.

Failure diagnosis

To evaluate the receive and transmit error registers of the individual communication participants ( 19 ... 24 ) in the control system ( 1 ), only the transfer of these register values in a special message is necessary. The control system ( 1 ) recognizes this message in the data traffic ( 3 ) and is thus able to recognize creeping participant failures at an early stage and to make corresponding parameter changes.

Load limitation

Due to the bus access of the CAN, a gush of a few, high priority Messages cause the waiting time of the remaining messages to increase extremely.  

In the structure according to the invention, precautions are taken to recognize the surge of these messages and to initiate suitable measures for limitation. These measures include a temporary increase in the gross data rate, the pause times t _Pi and the priority of the messages concerned.

Label list

1

Control system

2

industrial communication system

3rd

Traffic

4

Configuration command

5

technological process

6

actual requirements

7

operator

8th

Operator requirements

9

Operator observation

10th

Bus monitor with functional units for data stream analysis

11

Database

12th

Selective regulator

13

central control device

14

Connection from bus monitor to database

15

Connection from database to bus monitor

16

Connection from database to selective controller

17th

Connection of selective controller to central control device

18th

Connection of selective controller to database (FeedBack)

19th

. . .

24th

Communication participants

25th

Communication line

26

Coupling element

27

decentralized control device

28

Classification tables

29

Network structure table

30th

Traffic table

31

Participant table

32

Message table

33

static configuration table

34

dynamic requirements table

35

. . .

38

Connections from classification tables to static configuration tables

39

Connections from static configuration table to dynamic requirements table

40

Selector

41

Switching

42

. . .

47

Optimization strategies

48

left receiver

49

right receiver

50

Command receiver on the left

51

Command receiver on the right

52

Coupling element activation on the left

53

Coupling element activation on the right

54

upper reception register

55

lower receive register

56

top filter, memory, priority and actuator management

57

lower filter, memory, priority and actuator management

58

upper queue

59

lower queue

60

right transmitter

61

left transmitter

62

decentralized control device of the coupling element

Claims (5)

1.System for the process-adaptive optimization of industrial communication systems, characterized in that for online control of an industrial communication system which is determined in its communication-technical characteristics by the requirements of a technological process and selected and put into operation via the operator requirements ( 8 ) and corresponding start parameters ( 2 ) and constant adaptation of the same to the actual, operationally fluctuating requirements ( 6 ) of a technological process ( 5 ) a control system ( 1 ) is used to determine the communication-technically relevant characteristics of the industrial communication system ( 2 ) based on the data traffic ( 3 ) and issuing corresponding configuration commands ( 4 ) for changing the parameters and / or the structure of the industrial communication system ( 2 ) during operation by means of an opt contained in and selected by the control system ( 1 ) strategies ( 42 . , , 47 ). 2. System according to claim 1, characterized in that the control system ( 1 ) a bus monitor ( 10 ) for monitoring and analyzing the data traffic ( 3 ), a storage system for. B. in the form of a database ( 11 ) for storing communication-related characteristics, a selective controller ( 12 ) for determining the necessary parameter changes in the industrial communication system ( 2 ) and a central control device ( 13 ) for transmitting the parameter changes to the industrial communication system ( 2 ) . 3. System according to claim 1 and 2, characterized in that in the selective controller ( 12 ) on the basis of an upstream selector ( 40 ) one of the integrated optimization strategies ( 42 ... 47 ) is selected and activated, which determines the necessary parameter changes and sends them to the Central control device ( 13 ) for transmission by means of configuration commands ( 4 ) to the decentralized control devices ( 27 ) present in the communication participants ( 19... 24 ) of the industrial communication system ( 2 ), which recognize the required parameter change and carry out the corresponding control intervention. 4. System according to claim 3, characterized in that in the optimization strategies ( 42 ... 47 ) depending on the technological process ( 5 ) and the industrial communication system ( 2 ) used optimization interventions, which z. B. the change of the gross data rate, the allocation of priorities, the change of the bus access, the adaptation to discontinuous changes in requirements, the adaptive subscriber selection, the load limitation and the monitoring of the communication participants ( 19 ... 24 ) can be individually or in a suitable combination are included. 5. System according to claim 1 to 4, characterized in that in the industrial communication system ( 2 ) in addition to the communication participants ( 19... 24 ), which computers, controls, sensors and / or actuators can be and via communication lines ( 25 ) with each other Structuring elements, so-called coupling elements ( 26 ), are used for the process-dependent structure change of the industrial communication system ( 2 ) through network segmentation, network aggregation, network combination and / or network switchover to further communication lines ( 25 ), the structure changes being prompted by the control system ( 1 ) , analogous to the parameter changes and the coupling elements ( 26 ) contain two functional units for optimal adjustability, "filter, storage, priority and control device management" ( 56 and 57 ), which also include decentralized control devices ( 62 ) which, like the ( 27 ) of the communication participants ( 19 ... 24 ) can be adjusted to adapt to the process.