JPH02244203A - Control system and optimum property deciding device - Google Patents

Abstract

PURPOSE:To deal with even the new situations via the learning and to attain the flexible control with high performance by stereotyping the expertness of an operator and the waveform of the control result, judging the feature of the waveform, and performing the control according to the waveform feature. CONSTITUTION:A control subject 1 includes plural actuators and is controlled by an actuator controller G6. The actions of the subject 1 and those actuators are detected by a detecting device G14 and inputted to an optimum property deciding device G13. The device G13 recognizes plural input detector output signals as patterns and decides the manipulated variable of each actuator based on the degree of similarity between those input patterns and those stored previously. Then the device G13 generates the command signals to the controller G6 and a teaching device G16. Thus the manipulated variable of each actuator is decided via the fuzzy inference against the degree of confidence of each characteristic pattern. Then the satisfactory control performance is attained in the same way as performed by an expert operator.

Description

Detailed Description of the Invention [Industrial Field of Application] The present invention provides a control system for controlling a controlled object operated by a plurality of actuators, in which the operation of the controlled object and the plurality of actuators is comprehensively judged and one individual The present invention relates to a control system and an optimality determination device that determine the optimal control amount of an actuator.

[Conventional technology]

Conventional control devices perform control using multiple output signals that represent the operating status of the controlled object, but in a method that controls by looking at individual outputs, control is localized and the entire system is not controlled. This method has the disadvantage of not being able to consider optimality.

Therefore, recently there is a tendency to use a plurality of signals to find overall optimality. For example, the operation of the conventional control device and method will be explained using a rolling mill system that is complex and cannot be controlled by a single control device.

A rolling mill is a system that obtains a steel material of a desired thickness by controlling the distance between opposing rolls and the tension applied to the rolled material. However, flat steel materials cannot be obtained due to thermal deformation due to heat loss generated during rolling and roll deformation due to mechanical deformation, etc. Therefore, shape control has been developed to obtain flat properties.

However, the physical properties of rolling are affected by various factors.
Therefore, even if a control model is created and controlled near a specific operating point, the model will often be inconsistent with the actual operation of the rolling mill. For this reason, feedback control, which would produce good results if the model was accurate, could not fully demonstrate its capabilities, and there was a problem in that it could not outperform a skilled operator who operates based on intuition and experience.

[Problem to be solved by the invention]

The above-mentioned conventional technology does not take into consideration the point of making use of the know-how of a skilled operator, and there are gaps in control performance.

An object of the present invention is to provide a highly expandable control system and optimality determination device that incorporates the know-how of a skilled operator.

[Means to solve the problem]

The above objective is achieved by quantifying the intuition and experience of a skilled operator and determining the amount of operation of the actuator.

[Effect]

Skilled operators extract characteristic patterns from control variables and perform ambiguous operations. Similarly, the reliability of the characteristic pattern is obtained by calculating the sum of products of the control amount and passing the result through a nonlinear circuit, and the operation amount of the actuator is determined from the reliability of each characteristic pattern by fuzzy inference. Thereby, the control system operates like a skilled operator and good control performance can be obtained.

Further, even if the know-how of a skilled operator is stored as control knowledge and the control is performed using the knowledge, the same control performance as described above can be obtained.

〔Example〕

An embodiment of the control system of the present invention is shown in FIG. 1 below.

The controlled object 1 includes a plurality of actuators, and the actuators are controlled by an actuator control table [G6] that generates a control execution command for each actuator. The operation of the controlled object 1 and the plurality of actuators is determined by a device composed of various detectors arranged respectively.The plurality of detector output signals of the detection device G14 are input to the optimality determining device 1G13. The optimality determination device Gl3 includes an input unit that recognizes the plurality of input detector output signals as patterns, a memory function that stores the input patterns, and a plurality of patterns in which the input patterns are stored first. a processing unit that outputs the degree of similarity with the previously stored pattern; and a command generation mechanism 12 that determines the operation amount of each actuator from the degree of similarity and generates a command signal to the actuator control device G6. It is configured. In addition, in storing the pattern in the pattern storage function of the optimality determining device G13, the operator sets the optimal output for the input in advance and operates the optimality determining mechanism, so that the best manipulated variable is output. with the ability to recognize.

The structure is such that a teaching device G16 can be added to change the pattern storage contents in advance when the controlled object 1 or the actuator is changed.

On the other hand, the conventional system does not have the optimality determination device G13, and generally the actuator control is performed from the detection device G14.
! ! The system was configured to send the data to device G6 and optimize control for each individual actuator. For this reason, the optimization of the entire system could not be achieved, but as in the present invention, the optimality determination device [G
By adding 1 to 3, it is possible to determine the optimality of the entire system, and even if one of the actuators fails, this determination device can cover the function of the failed actuator with another actuator, or replace the actuator. It has the effect of being able to flexibly respond to changes and changes in the controlled object.

Next, a detailed explanation will be given using an example in which the control system of the present invention is applied to rolling mill control.

An embodiment in which the present invention is applied to a rolling mill control system will be described below with reference to FIG.

The rolled material stand 1 to be controlled crushes and pulls the rolled material 3 sandwiched between a pair of opposing work rolls 2 by the rolling force acting between the work rolls 2 and the tension acting on the rolled material 3. The rolled material is thinned by force to obtain a desired thickness, and an intermediate roll 4 is arranged to sandwich the work roll 2, and a backup roll 5 is arranged to sandwich the intermediate roll 4. A rolling force is applied to the backup roll 5 by a rolling down control mechanism 6 using force such as hydraulic pressure, and the rolling force is transmitted to the intermediate roll 4 via the contact surface between the backup roll 5 and the intermediate roll 4, The rolling force transmitted to the intermediate roll 4 is transmitted to the rolled material 3 through the contact surfaces between the intermediate roll 4 and the work roll 2, and between the work roll 2 and the rolled material 3, and the rolling force causes the rolled material 3 to become plastic. Deformation occurs and the desired thickness is achieved.

By the way, the roll width of the rolling rolls 2 and 4.5 is the same as that of the rolling material 3.
Because the width of the roll is wider than the width of the roll and rolling force is applied, the roll deforms. For example, a portion of the work roll 2 that is outside the width of the rolled material is bent by the rolling force.

As a result, the ends of the rolled material 3 are crushed and have a convex cross-sectional shape. To prevent this, for the axis of work roll 2,
A work roll pending force Fw is applied by the work roll bender 7 in the direction in which the interval widens, thereby preventing the ends of the rolled material 3 from being crushed. Similarly, an intermediate roll pending force F+ is applied to the shaft of the intermediate roll 5 by an intermediate roll bender 8.

Further, the intermediate roll shift 9 moves the intermediate roll 4 to the plate 11]
move in the direction. This movement results in a roll of 2.

By making the forces applied to 4, 5 and the rolled material 3 asymmetrical, the shape of the plate thickness of the rolled material is controlled.

On the other hand, the energy applied to the rolling mill for rolling is not only used for plastic deformation of the rolled material 3, but also for noise, vibration, etc.
It becomes fever. This energy converted into heat is dissipated through the rolled material 3 and increases the temperature of the work roll 2. Due to this temperature increase, the work roll expands and the roll diameter changes, but the roll diameter generally deforms non-uniformly. Therefore, in order to control the roll diameter uniformly,
A plurality of nozzles (not shown) arranged in the width direction of the plate and a coolant control mechanism 10 that applies cooling liquid to the work roll 2 through the nozzles are installed.

A speed control mechanism 11 composed of an electric motor or the like for moving the rolled material 2 is connected to the shaft of the work roll 2 .

The control method for the rolling mill is as follows: the rolling control mechanism 6°, the work roll bender 7, the intermediate roll bender 8. Intermediate roll shifter9. Coolant control mechanism 10. A command generating mechanism 12 that generates operation commands for actuators such as the speed control mechanism 11. A pattern recognition mechanism 13 for determining, for the command generation mechanism 12, which type of pattern the shape of the rolled material 3 belongs to among a plurality of pre-stored patterns, and outputting the certainty factor of the pattern. The pattern recognition mechanism 1
3, a shape detection mechanism 14 that detects and outputs the plate thickness shape of the rolled material 3, a memory mechanism 15 that stores the outputs of the shape detection mechanism 14 and the command generation mechanism 12, and a memory mechanism 1.
The learning mechanism 16 is configured to change the parameters of the pattern recognition mechanism 13 through learning using the information of No. 5.

FIG. 3 shows a detailed view of the shape pattern recognition mechanism 13. The outputs of the shape detection mechanism 14 and the storage mechanism 15 are input to input cells 17 and 18 of the pattern recognition mechanism 13,
In the human cell 17, the input signal is converted into a function value and output to the intermediate layer 19, and the output of the input cell input to the intermediate layer 19 is input to the cell 20.21 of the intermediate layer 19.

The signal inputted to the cell 20 as the output of the input cell 17 is multiplied by wzt by the weighting function 23 and inputted to the adder 24, and the output of the input cell 18 is inputted to the adder 24 via the weighting function 26 and added. The weight function 23.26
The sum of the outputs is added and input to the function unit 258, where the function unit 25 performs a linear or non-linear function operation, and output to the next stage intermediate layer 27. Note that the cell 20 has the above weighting functions 23.26.
It is composed of an adder 24 and a function unit 25.

Similarly, the output of the input layer 18 is input to the cell 21, and the output of the input layer 17 is multiplied by w12 by the weighting function 28 and added to the adder 29. It is output to the next stage intermediate layer 27 via the function unit 30.

The intermediate layer 27 has the same structure as the intermediate layer 19, and the output of the intermediate layer 19 is used instead of the output of the input layer 17.18.

Here, the weight to be applied to the i-th output of the k-1th intermediate layer (input cell when k=1) in the cell of the weighting function 23, 26.28 is shown.

The signals input to the pattern recognition mechanism 13 as described above are transmitted to the input cells 17, 18, . Multi-level intermediate layer 19.27.2
9, and an output layer 30 which is obtained by removing the weighting function and adder from the cells of the intermediate layer. Note that the input layer 31 represents a collection of all the human cells 17 and 18.

The characteristics of this pattern recognition mechanism 13 are that it only requires a simple product-sum operation, and there is no repetitive operation such as feedback, and that each product-sum term in the intermediate layer can be processed in parallel when implemented in hardware. High-speed calculation is possible.

It is also possible to store command values for each actuator in accordance with each output pattern in advance in the output layer 30 of this pattern recognition mechanism, and to directly instruct the actuator with the command value closest to the output pattern. Although this method has good responsiveness, the control accuracy is slightly worse than the method described below.

Next, the processing result of the pattern recognition mechanism 13 is applied to the rolling mill 1 via the processing mechanism shown in FIG. That is, the output of the pattern recognition mechanism 13 is input to the operation amount determining means 32 provided in the command generation mechanism 12. The manipulated variable determining means 32 selects the most effective processing mechanism for processing the input signal from among the plurality of internally prepared processing mechanisms, executes the process, and outputs the manipulated variable. Using the results of the operation amount determining means, the command value calculating means generates specific command values 1 for each actuator, such as a rolling down command for the rolling down control mechanism 6, an intermediate roll bender command for the intermediate roll bender 8, etc. Note that the output of the shape detection mechanism 14 is directly input to the command generation mechanism 12 without going through the pattern recognition mechanism 13, and is processed by the optimal processing mechanism among the plurality of processing mechanisms prepared inside. It is also possible. However, in this case, when using various inference mechanisms, it is necessary to enrich the knowledge base in order to sufficiently reflect the operating methods of specialized operators.

FIG. 5 shows the configuration of the operation amount determining means 32. The operation amount determining means 32 includes the shape detection mechanism 14. Upon receiving a signal from the pattern recognition mechanism 13, the control mechanism 141 is activated. The control mechanism 141 operates according to the type of one problem.
Using Knowledge Base 36.

Decide which inference to invoke. That is, the control mechanism 141 activates the production inference mechanism 142 when it is necessary to find a cause using a syllogism, and activates the fuzzy inference mechanism 143 when there is an ambiguous factor, and solves problems that have a certain framework. For problems, the frame inference mechanism 144 is activated, and the semantic net inference mechanism 145 is activated for problems in which causal relationships, device configurations, etc. are related in a network manner. The script inference mechanism 146 is activated for a problem that appears to be working. Furthermore, the control mechanism 141 is an empirical problem that cannot be solved by the various reasoning mechanisms described above. Activates the optimization calculation mechanism 111 to find the optimal solution at high speed, and the feature extraction/answer mechanism 110 (Ruen Alhart type neuron) that can be memorized in patterns, extracts features, and solves problems that require answers. (configured on your computer). Operation amount determining means 32
The processing results are output to the command value calculation mechanism via the control mechanism 141.

FIG. 6 shows the configuration of a knowledge base 36 that is knowledge necessary for inference. The knowledge base is based on the experience of a control expert, etc. Knowledge 106 input from the outside is a product for performing inference using a syllogistic method. Rules 147, Fuzzy rules 148, which are knowledge for making inferences based on ambiguous information, Knowledge frames 1499, which can be described in a certain framework, such as the configuration of parts to be diagnosed, and the relationships between parts and common sense relationships. Net2 91509 A script 151 that organizes and stores certain tasks when the diagnosis target performs them in order, and other knowledge that cannot be described in the above knowledge 147 to 153. It is classified into 54 categories and stored.

FIG. 7 shows an explanatory diagram of the operation of the manipulated variable determining means 32. The processing of the control mechanism 141 is performed by the pattern recognition mechanism 13, shape detection mechanism 14. a processing step 200 for organizing information from storage 15 and converting it into data that can be used for further processing;
Repetitive processing step 201 that retrieves the data prepared in step 200 above and passes it to step 202 until it is exhausted.
, a judgment step 202 for determining the inference mechanism and process to be activated from the information collected in step 201;
and various inference mechanisms 142 to 146, feature extraction and response mechanism 110, optimization calculation mechanism 111, and general control mechanism 203 that executes algorithms of classical control such as PID control and modern control such as multivariable control; It consists of an end processing step 204 that executes reset of flags etc. necessary to end each step.

Here, the role of each processing mechanism will be described. The production mechanism 142 is suitable for control in which an expert operator assembles a logically established relationship using fragmentary production rules. Fuzzy reasoning 143 determines the amount of operation by quantifying the operator's ambiguous knowledge that cannot be quantified, such as the operator moving the actuator slightly if the state of interest of the controlled object changes, so that it can be processed by a computer. suitable for.

The frame inference mechanism 144 uses the knowledge of frames that describe relationships between control devices, and performs operations based on the relationships between these devices when returning to the original state when the state of the controlled object of interest changes. It is suitable for determining the amount of processing to be performed for each related device.

The semantic net inference mechanism 145 organizes the frames, which are the fragmentary knowledge, and organizes them into a system to create a network, so it can determine the influence of the operation result of a specific actuator and create a compensation system. suitable for.

Since the script inference mechanism 146 makes inferences based on procedural knowledge when a specific state occurs, it is suitable for sequence control-like control where a fixed procedure must be followed in the event of a failure or the like.

The feature extraction/answer mechanism 110 also includes the pattern recognition mechanism 13. Shape detection mechanism 14. If the relationship between the input pattern of the storage mechanism 15 and the output produced by the inference mechanisms 142 to 146 when the input pattern is input is learned in advance, the inference mechanisms 142 to 146 will perform inference and determine the output. Unlike , it has the feature that it can output the same results at high speed. The optimization calculation mechanism 111 normally has strong nonlinearity in the controlled object 1, and if the operating point changes for some reason, the operation needs to be reset.

Linear programming, hill climbing method, conjugate slope method or Ho
It is calculated by an algorithm such as a pfiald type neurocomputer, and provides an optimal response even to a nonlinear controlled object.

FIG. 8 shows an explanatory diagram of the operation of the production mechanism 142. The production inference mechanism 142 activated by the control mechanism 141 retrieves from the control mechanism 141 the input process 34 stored in the memory at the time of activation, and the information stored in the input process 34 one by one, and if there is no pattern information in the memory. At times, the termination determination mechanism 35 is executed to terminate the processing of the production inference mechanism 142. Using the pattern type and its certainty factor extracted by the termination determination mechanism 35, rules are extracted one by one from the knowledge base 36, and in process 37, the input pattern type and the antecedent part of the rule are compared. Using the comparison results, step 38 executes the next process 39 if they match, and executes step 37 if they do not match. Step 39 replaces the input with the conclusion part of the rule when there is a match. At this time, the confidence level is handled using the Minimax theory, and is replaced with the minimum value or maximum value before replacement. In step 40, if the conclusion part of the replaced rule is an operation command, step 41 is executed, and if the conclusion part does not match, step 37 is executed to further perform inference.

When the conclusion part is an operation command, the process 41 outputs the conclusion part and the confidence obtained in the process step to the command value calculation means 33.

FIG. 9 shows the command calculation means 33. Command value calculation means 33
A memory 42. is a memory 42 for storing commands and their reliability which are the inference results obtained by the operation amount determining means 32. A step 43 in which it is determined whether all the commands in the memory have been processed, and if they have been processed, the command value calculation means 33 is terminated; if not, the actuator 7 of the lowering control mechanism 6, etc.
．． 8゜9.10.11 Extracting the commands and based on the degree and confidence of actuator operation determined by various inferences,
It consists of a process 44 in which the center of gravity of the manipulated variable is determined, the center of gravity of the manipulated variables of the same actuator is gathered together, a new center of gravity is determined, and the center of gravity is used as a command for the corresponding actuator.

By providing such a command value calculation means 33, various inferences 142 to 146. Feature extraction/circuit mechanism 110, optimization calculation mechanism 111. - It has the feature that commands to the actuators individually obtained by the membrane control mechanism 203 can be handled in a unified manner.

FIG. 10 shows an input switching device [125
The configuration is shown below. The input switching device 1i125 outputs either the output of the shape detection mechanism 14 or the output of the learning mechanism 16 to the input layer 31 using a switch mechanism 156 controlled by a learning mechanism.

The state of the switch mechanism 156 in FIG. 10 indicates a state in which learning is performed.

FIG. 11 shows the configuration of the learning mechanism 16. The learning mechanism 16 includes an input pattern generating mechanism 45. Output pattern generation mechanism 4
7. It is composed of an output matching mechanism 46 and a learning control mechanism 48. The output matching mechanism 46 outputs the outputs o1 and HOn of the distributor 139 for outputting the output of the output layer 30 to the command generation mechanism 12 and the matching mechanism 46, and the output OTI HOTI HO of the output pattern generation mechanism 47.
Adders 161, 162, and 163 add the difference from Tn to the deviation el.

eJ and en are obtained and output to the learning control mechanism 48.

Note that the output ol of the distributor 139 is 01.0. The output of the input pattern generation mechanism 47 is the pattern recognition mechanism 13 (Ru
Melhart type neurocomputer) input layer 19
Occurs when input to . At this time, the input pattern generation mechanism 45 and the output pattern generation mechanism 47 are controlled by the learning control mechanism 48.

Figure 12 shows the weight function W I J2 in the learning process.
3 and the learning control mechanism 48 are shown. The adder 161
In response to the deviation ek, which is the output of
The value of J23 is changed in a direction that reduces the deviation.

FIG. 13 shows a processing overview 170 of the learning control mechanism 48. When the learning mechanism 16 is activated, a process 170 of the learning control mechanism 48 is activated. The processing 170 includes the input pattern generation mechanism 45. Preprocessing 171. starts the output pattern generation mechanism 47 and generates an input as a teacher signal and a desired output. Step 172. Repeat steps 173, 174 and 175 until the value of the deviation ek or the sum of the squares of the deviations falls within the allowable range. A step 174 of sequentially extracting the target intermediate layer from the intermediate layer close to the output layer 30 toward the input layer 31, a step 174 of sequentially extracting the target cells in the intermediate layer, and cells extracted in the direction where the deviation ek becomes smaller. It consists of step 175 for changing the weight function WIJ23 of , and step 176 for terminating the learning process.

By providing such a learning mechanism, a new phenomenon that has not been considered until now occurs, and once a countermeasure against it has been determined, it has the characteristic of being able to reflect that knowledge.

FIG. 14 shows the configuration of the storage mechanism 15 of FIG. 2. The storage mechanism 15 includes the command generation mechanism 12. Memory element 49 to which the output of the shape detection mechanism 14 is input; the contents of the memory element 49 are transferred from the memory element 5o, which is transferred after a certain period of time, and from the memory element 51, which is sequentially transferred to the memory elements and arrives after a certain period of time has elapsed. configured, each memory element 49,5
The contents of 0.51 are sent to the pattern recognition mechanism 13. The information is input to the learning mechanism 16.

This storage mechanism 15 enables operations such as differentiation and integration, which can take into account temporal changes in the shape detection mechanism 14 and the command generation mechanism 12, to be performed.

FIG. 15 shows a mechanism for recognizing a pattern using an input near the nozzle because the influence of the nozzle in coolant control affects only a certain length from the nozzle position. The output of the shape detection mechanism 14 is sent to the memory 52 of the pattern detection mechanism 13.
The signal input to the memory 52 is input to the memory element 54 via the gate circuit 53, and the signal is input to the memory element 54.
The signal input to 4 is input to memory element 57.58 via gate circuit 55.56, and when gate circuit 53. to the memory 57, and
After a certain period of time, the signal from memory element 54 reaches memory element 58, and the signal from memory element 57 reaches memory element 54.
, and when the signals in memory elements 54, 57, and 58 complete one cycle at the next clock, gates 53, 56 are turned on, gates 55 are turned off, and the contents of memory element 54 are transferred to memory element 59.
The information in the memory elements 54, 57, and 59 is input to the input layer 31. .

By providing such a memory 52, the input layer 31 . Middle layer 19°27.29. This has the effect of reducing the number of cells in the output layer 30 to large + IJ.

FIG. 16 shows an input pattern generation mechanism 45 of the learning mechanism 16, an output pattern generation mechanism 47, and a controlled object simulator 60.
Here is an example using .

The shape pattern generated by the operator ψ operation or data in the output pattern generation mechanism 47 is input to the command generation mechanism 12, which has the same function as the command generation mechanism 12 shown in FIG. Mechanism 1
2, commands for various actuators are generated according to the pattern, and the commands are input to the controlled object simulator 5o provided in the input pattern generation mechanism 45, and the various actuators 6.7, 8, 9, 10 degrees that are the controlled objects are generated. 11 and the rolling mill 1, and when the response is poor, the command generation mechanism 12. A parameter adjustment mechanism 51 for changing parameters of the controlled object simulator 50 is used to adjust the output of the controlled object simulator 50 to a desired shape, and the output is input to the pattern recognition mechanism 13.

The operation of the control method having the configuration described above will be described below using a specific example.

The initial value of the weighting function Wtj28 of the intermediate layer 19, 27, 29 of the neurocomputer configuring the pattern recognition mechanism 13 is initially a random number or an appropriate value, for example, a value that the weighting function can take (assuming O to 1.0). ) half (0,5)
Set to . At this time, for example, even if the concave rolling mill shape pattern generated by the delivery input pattern generation mechanism 45 in FIG. Also, output layer 30
The probability that the output of the output line 71 is convex does not become zero.

Therefore, the learning mechanism 16 corresponding to the output line 70 of the output layer 30
The output line 72 of the output pattern generation mechanism 47 is connected to 1, and the output line 7 of the output pattern generation mechanism 47 corresponding to the output line 71 is
Output the output of 3 to zero.

In response to these outputs, the output matching mechanism 46 receives the associative output (output pattern generation mechanism 47) and the deviation between the output of the pattern recognition mechanism 13.The learning control mechanism 48 receives the deviation of the output of the pattern recognition mechanism 13. The magnitude of wiJ is changed in a direction in which the deviation decreases in proportion to the magnitude of the deviation. A typical example of this algorithm is the steepest slope method.

The weight of the weight function is sequentially changed according to the process shown in FIG. 13, and when the sum of squares of ek shown in FIG. 12 falls within the allowable range, the operation of the learning mechanism 16 ends.

After learning is completed, when the same waveform as the output pattern of the input pattern generation mechanism 45 in FIG. 17 is input from the shape detection mechanism 14 in FIG. 2, the pattern recognition mechanism 13 outputs 1 from the output line 7o of the output layer 30. Then, zero is output from the output line 71 of the output layer 30.

Next, if the waveform shown in FIG. 18, which is said to be convex, is input, and learning has not yet been completed, the output of the output line 71 representing the convex shape of the pattern recognition mechanism 13 is 1, and other Therefore, as described above, by using a typical convex pattern as an input signal, the output of the output pattern generating mechanism 47 is connected to the output line 71.
．． The values corresponding to the outputs of 70 are set to 1 and 0, respectively.

The learning mechanism 16 changes the weight function W I J, and when the learning is completed, changes the weight function W I J to the pattern recognition mechanism 13.

When the convex waveform shown in FIG. 18 is input, the output line 71 of the output layer 30 shown in FIG. 17 becomes 1, and the output line 70 becomes 0.

As a result, the waveform shown in FIG. 19(a) is the pattern recognition mechanism 1.
3, and its output is the convex waveform inputted in advance from the output layer 30 as described above.
1, the degree of similarity to that waveform is output with a certainty of 40%, and at the same time, the degree of similarity to the waveform is output with a certainty of 50% from the output line 70 indicating that it is a concave waveform.

Fig. 20 shows the shape of the rolled material taking into consideration the temporal changes in the rolled material.The state directly below the rolling mill 2 is to,
The value at that time is xo. If the sampling period of the calculator is To, the height of the plate thickness at time t1, To seconds ago, is XI, and the height of the plate thickness at time tz, Toxz seconds ago, is XZ, etc. .

That is, at time tz, the height xz is input to the storage mechanism 15, and the height xl of tl, which is the 0th-order sampling point stored in the memory element 49 of FIG. 14, is input to the storage mechanism 15. At that timing, the data x2 in the memory element 49 is transferred to the memory element 50, and the contents of the memory element 49 are rewritten to xl.

On the other hand, the calculation mechanism 510 performs various calculations using the contents of the memory elements 49 and 50. For example, when you need a differential value, you can use (xz-xt)/To, and when you need an integrator, you can use (x1+xz)XTo.
That is, since the differentiator can determine the rate of change in shape, the pattern recognition mechanism 13 can improve responsiveness to changes.

On the other hand, an integrator can exhibit characteristics such as having a noise removing effect.

The pattern recognition mechanism 13 can have functions such as a differentiator, an integrator, and a proportional element that does not include a temporal element.

Furthermore, the data stored in the storage mechanism 15 can also be stored as necessary.
It can be used in the input pattern generation mechanism 45 used during learning.

By the way, as shown in FIG. 29, the state of the thickness of the rolled material in the roll axis direction of the rolling mill at the time "to" before To (sampling period) is then stored in the memory element 54 of FIG. At .57.58, the main cable 59 performs the same operation as the memory mechanism 15 described above, so the data is stored at time 11, which is before time To.

FIG. 22 shows an example of a production rule or fuzzy rule (corresponding to production rule 47 and fuzzy rule 48 in FIG. 6).

When the pattern recognition mechanism 13 obtains an output as a concave 50% confidence level, it is compared with the premise of the production rule and matches the concave rule 80.

As a result, a rule 81 that weakens the vendor (Small) is obtained. On the other hand, with confidence of convex type 40%, 5 premises 8
2, and as a result, the vendor is strengthened (to a large degree).

As a result, as shown in FIG. 23, the command generation mechanism 12:
As a result of the comparison with the rule, the vendor's operation amount has a convex shape with a certainty of 50%, so it is represented by the area of the diagonal line in B. on the other hand,
Since the confidence of concave shape is 40% and the confidence of S is 40%,
This is the area of the shaded part S in FIG. 22. Next, in the command generating mechanism 12, 65%, which is the value of the center of gravity C which is a combination of the centers of gravity A and B in the shaded area, becomes the amount of operation by the bender.

Next, in cases where the influence of the actuator is local, unlike that of the bender or shifter, such as coolant control, the waveform of FIG. 24(a) is transferred to the memory element 54 as shown in FIG.
57, 58. A part of the waveform stored in the memory element (the part marked ■ in FIG. 24(a)) is transmitted to the pattern recognition mechanism 13. This is processed by the command generation mechanism 12, and by controlling one nozzle A of the coolant control device 10, the amount of coolant is controlled, and the roll is flattened.

Now, if Xa-1 in FIG. 21 corresponding to nozzle A is large, the conclusion 85 that the center is large becomes the 15th, so the differential coefficient becomes positive, matching the premise 86, and as a result, the coolant is turned on. do. It will stop changing to that extent.

When the control of nozzle A is completed, the memory elements 54 and 54 of FIG.
The contents of 57, 58, and 59 are shifted by one. As a result, the waveform input to the pattern recognition mechanism 13 is in the area indicated by ■ in FIG. 24(a), and the processing 13.
12, one nozzle B of the coolant control mechanism 10 is controlled.

When the processing is performed in this manner, starting from the pattern (2) in FIG. 24(a) and arriving at pattern (2), and further shifting the memory contents, the waveform shown in FIG.

After a certain period of time has elapsed since the pattern shown in FIG. 24(a) was previously stored in the memory 52, the contents of the memory element 54 shown in FIG. let

Furthermore, it is obvious from FIG. 14 that if the arithmetic mechanism 510 shown in FIG. 14 is provided between the memory 52 and the input layer 31, it becomes possible to control the changing speed of the waveform, etc.

Next, a method for learning a reference pattern for the pattern recognition mechanism 13 will be described.

Waveforms 62 and 63 in FIG. 18 are generated by the input pattern generation mechanism 45 in FIG. 11 and output to the input layer 31. This pattern is written into the memory of the input pattern generation mechanism 45, or a pattern stored in the storage mechanism 15 of FIG. 2 is used. The signal input to the input layer is transmitted to the intermediate layer 19.・・・
. , 27 and appears as an output from the output layer 30.

At this time, the weight function ω1 of the intermediate layer. is the initial value, and from the output pattern generation mechanism 47, the input pattern generation mechanism 45
Corresponding to the output of 1 output terminal
, and the other terminals become zero) is input to the matching mechanism 46. When learning is not completed,
Output pattern of output layer 30 and output pattern generation mechanism 4
7 have different waveforms. As a result, the butting mechanism 46
The output corresponds to the degree of pattern difference. By calculating the square mean of this value and deviation, the power spectrum of the deviation etc. can be obtained. Various methods can be considered for changing the 0 weighting function WsJ, which changes the 1 weighting function w from the intermediate layer 27 close to the output layer to the intermediate layer 19 close to the input layer 31, in accordance with the above deviation. This is an optimization problem in which the above deviation is made to be the minimum value, and for example, the steepest slope method is used. As a specific method, the weighting function W I
When J is slightly changed upward, looking at the direction in which the deviation value changes, the value of the weighting function W I is changed in the direction in which it decreases.
J is moved, and the amount of movement is large when the change in the deviation value is small, and conversely, the amount of movement is made small when the change in the deviation value is large. Furthermore, when the change of the weighting function wIJ of the intermediate layer 19 closest to the input layer is completed, the deviation value of the matching mechanism 46 is checked again, and when the value falls within the allowable error range, learning is completed.

This control is performed by a learning control mechanism 48. However, it is not clear why the pattern recognition mechanism 13, which uses the learned results for pattern discrimination, is able to identify patterns or why the learning is successful. It is said that the number of patterns is large compared to the number of patterns, and there is a degree of freedom in the values, and even if the values are slightly out of order, or even if many patterns are memorized, it is possible to obtain good recognition results.

On the other hand, it is very difficult to determine what kind of patterns should be used for the input pattern generation mechanism 45 and the output pattern generation mechanism 47. Fortunately, a theory called system identification has been established in the field of control theory that allows a model to be derived accurately by operating the controlled object 1 near a certain operating point. However, it is difficult to model the object with strong nonlinearity in the entire operating region.

Therefore, we create a model in a specific operating region, implement control, use simulation to determine the relationship between the input and response of the control system that works well under that condition, and use this as data for learning. This procedure is learned by sequentially moving the operating points over the entire operating range of the control system and determining the optimal modeling and control commands at each time. That is, the controlled object simulator 6 in FIG.
0 parameters are adjusted to cause the simulator 60 to operate accurately at a specific operating point. After that, the input pattern generating mechanism 47 . Parameter adjustment mechanism 61. Controlled object simulator 60. The command generating mechanism 12 is operated, and the outputs of these processes 47 and 60 are used as the output pattern and input pattern of the learning mechanism 16, respectively.

In a control system with such a configuration, the pattern recognition mechanism abstracts the target waveform, and the control mechanism can perform control including ambiguity.

Although the embodiment has been described using a rolling mill system as a specific example of the present invention, the control object 1. It is obvious that the various actuators 6.7, 8, 9, and 10.11 are not limited to rolling mill systems, and can be applied to general control objects, actuators, and controllers. It can be used to control a system that recognizes train schedule patterns and returns delayed trains to normal schedules according to various schedule change rules. In other words, the train operation is represented by a diagram, and the pattern:! The late features are extracted by the recognition mechanism 13. Next, based on the feature amounts, the inference mechanism creates a timetable using various rules such as, for example, passing trains at stations. In response to the results of the inference mechanism, the command calculation mechanism 33 generates operating commands for individual trains. It is an actuator. The train is
Operate according to the above instructions.

〔Effect of the invention〕

According to the present invention, in a pattern control method such as the shape control of a rolling mill, the operator expert categorizes the waveform of the control result, determines the characteristics of the waveform, and performs control according to the characteristics. It is possible to implement a system that allows for flexible and high-performance control even if a new situation occurs, because it can be dealt with through learning.

[Brief explanation of drawings]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is an embodiment in which the present invention is applied to a rolling mill control system, Fig. 3 is a pattern recognition mechanism diagram, Fig. 4 is a command generation mechanism diagram, and Fig. 4 is a diagram of a command generation mechanism. Figure 5 is a configuration diagram of the operation amount determining means, Figure 6 is a knowledge base configuration diagram, and Figure 7 is a diagram of the knowledge base configuration.
Figure 8 is an explanatory diagram of the operation of the operation amount determining means, Figure 8 is an explanatory diagram of the operation of the production mechanism, Figure 9 is a configuration diagram of the command value calculation means, Figure 10 is the configuration of the input switching device, and Figure 11 is the learning mechanism. 12 is a diagram of the relationship between the learning control mechanism and node weight functions, and FIG. 13 is a basic processing diagram of the learning control mechanism.
FIG. 14 is a configuration diagram of the storage mechanism, FIG. 15 is a diagram of the pattern recognition mechanism, FIG. 16 is a configuration diagram when the learning mechanism is equipped with a simulator, and FIG. 17 is an explanatory diagram of the operation of the pattern recognition mechanism.
Fig. 18 is an example of an input pattern, Fig. 19 is an explanatory diagram of the output of the pattern recognition mechanism, Figs. 20 and 21 are explanatory diagrams of temporal changes in rolled material, and Fig. 22 is an example of production rules and fuzzy rules. The diagrams shown in FIG. 23 are explanatory diagrams of a method of converting similarity into manipulated variables, and FIG. 24 is an explanatory diagram of the processing status of input waveforms. l...Controlled object, 13...Pattern L! ! intelligence organization,
12 Command generation mechanism, 16...Learning mechanism, 19, 27° Fig. 2 Fig. 1 Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Figure fig fig fig fig fig fig fig fig fig fig fig fig fig fig fig fig.

Claims (18)

[Claims] 1. A control system consisting of a control object operated by a plurality of actuators, a plurality of actuator control mechanisms that individually control the plurality of actuators, and a detection device comprising a plurality of detectors that detect operations of the control object and the actuator. , comprising an optimality determination device that comprehensively determines the operation of the control target and the plurality of actuators from the plurality of output signals of the detection device, and transmits a control command to the plurality of actuator control mechanisms based on the determination result. A control system characterized by: 2. In claim 1, the optimality determination mechanism comprises:
a control mechanism that determines a control strategy to be used for processing from a plurality of detection signals output from the detection device, and generates a processing command;
A control system comprising command value calculating means for converting the manipulated variable determined by the manipulated variable determining means into a control command for each actuator. 3. In claim 2, the plurality of control strategies stored in the manipulated variable determining means include an inference control strategy composed of a knowledge base and an inference section, a control strategy using a pattern recognition method, a feedback control, etc. A control system characterized by comprising either a control strategy using a compensator or a control strategy using a state model. 4. In claim 1, the optimality determination mechanism comprises:
means for storing a plurality of combination patterns of the plurality of detection signals of the detection device and operation amounts of the actuator corresponding to the plurality of combination patterns; A control system characterized in that it is configured to compare stored output patterns and output operation amounts of a plurality of actuators for the stored pattern with the greatest degree of similarity. 5. In claim 1, the optimality determination mechanism comprises:
a pattern recognition mechanism that determines the degree of similarity between the patterns of the output signals from a combination of a plurality of detection signals from the detection device; A control device comprising a command generation mechanism that converts command signals to a plurality of actuators. 6. In claim 5, the pattern recognition mechanism includes an input layer that takes in a combination pattern of a plurality of output signals of the detection device, and a function that multiplies and adds the output signals of the input layer with a weight, and uses the result as a designated function. The output signals of each node of the first intermediate layer are input signals, multiplied by weights, and added, and the results are mapped by a specified function. A control characterized in that the control has a plurality of stages of another intermediate layer composed of a plurality of nodes, and a final stage of the another intermediate layer is composed of an output layer that outputs pattern similarity as a determination result. system. 7. In claim 6, the command generation mechanism uses a knowledge base and an inference mechanism to determine the amount of operation for the actuator from the similarity of the plurality of patterns output from the pattern generation mechanism, and calculates the amount of operation for the actuator from the similarity of the plurality of patterns output from the pattern generation mechanism, and A control system characterized by being configured to convert it into a command signal for a control mechanism of an actuator. 8. Claim 1: A teaching device comprising: a mechanism for determining whether the determination result of the optimality determination device is good; a means for notifying the determination result to the outside; and a means for changing the contents of the optimality determination device. A control system characterized by: 9. In claim 1, a storage mechanism is provided for storing the output signals of the detection device in a chronological order, and the output of the detector and the output of the storage mechanism are input to the optimality determination device, and the optimality determination device A control system characterized in that the determination device is capable of comprehensive determination that also takes into account temporal changes. 10. In the control system, the control system includes a plurality of actuator control mechanisms that execute control according to detection signals of a plurality of detectors and output signals of the detectors, and a controlled object controlled by the operation of the plurality of actuators. Pattern recognition that recognizes a combination of multiple detection signals as a pattern, stores the signal combination pattern in advance, compares the pre-stored pattern with a new input pattern, and outputs the comparison result as pattern similarity. A control system comprising: a mechanism; and a command generation mechanism that converts the degree of similarity of the patterns into an operation amount for each actuator by fuzzy inference, and converts the operation amount into a command value for each actuator. 11. 11. The control system according to claim 10, wherein a Rummelhardt-type neurocomputer is used as the pattern recognition mechanism. 12. Claim 11, further comprising a learning mechanism for applying an input pattern to the pattern recognition mechanism and changing the weight of each node of the pattern recognition mechanism in advance so that the pattern recognition mechanism has an ideal output. A control system featuring: 13. In claim 11, the learning mechanism comprises: an input pattern generation mechanism inputting to an input layer of a pattern recognition mechanism; an output pattern generation mechanism generating an ideal output of the pattern recognition mechanism; an output pattern of the pattern recognition mechanism; a comparison mechanism for determining the deviation of the output pattern generation mechanism;
a command to change the weight of a node in the pattern generation mechanism based on the comparison result; and the input pattern generation mechanism;
A control system comprising a learning control mechanism that generates operation commands for an output pattern generation mechanism. 14. According to claim 10, the configuration includes a storage mechanism that stores the detection signals of the plurality of detectors in a chronological order, and inputs the output of the storage mechanism to the pattern recognition mechanism in parallel with the detection signals of the detectors. A control system characterized by: 15. A controlled object that operates with multiple actuators,
In a control system comprising a detector that detects the operation of the plurality of actuators and the operation of the controlled object, the operation of the entire control system is recognized using the plurality of detection signals so that the operation of the entire system is optimized. A control system comprising: an optimality determination device that determines a control amount for each of the actuators. 16. In a control system for a rolling mill that includes a plurality of drive mechanisms and rolls a material, detecting the states of the plurality of drive mechanisms and the state of the rolled material, and determining the quality of the operation of the entire system based on the detection signal, A control system configured to output a command signal to each of the drive mechanisms by classifying the amount of operation of each of the drive mechanisms. 17. It has a plurality of input terminals into which the operating states of a plurality of actuators can be input and a plurality of output terminals which output control commands to the plurality of actuators, and based on the input signals, determines whether the operation of the entire system is good or bad, and determines the determination result. Optimality determination device with a function to determine the output signal to each output terminal based on 18. According to claim 17, the system has a function of patterning a combination of operation signals of the entire system and storing said pattern, and determining a degree of similarity between a plurality of stored patterns and a pattern of a new input signal, and said degree of similarity. and an inference function for determining each output signal from the optimality determination device.