CN114637194A

CN114637194A - Plant control system, plant control method, and computer-readable recording medium

Info

Publication number: CN114637194A
Application number: CN202111242094.XA
Authority: CN
Inventors: 高田敬规; 服部哲; 阿部隆; 田内佑树; 黑川大辉; 绵岛正刚
Original assignee: Hitachi Ltd
Current assignee: Hitachi Ltd
Priority date: 2020-12-16
Filing date: 2021-10-25
Publication date: 2022-06-17
Also published as: DE102021214272A1; JP7495874B2; JP2022095082A

Abstract

The invention provides a plant equipment control system, a control method and a computer-readable recording medium. In a plant control system, a control rule is efficiently corrected with little risk of disturbing the control of plant equipment. The disclosed device is provided with: a control method learning unit that learns a combination of actual performance data and a control operation of the target plant equipment; a control execution unit that executes control of the target plant equipment based on a combination of the actual result data learned by the control method learning unit and the control operation; and a goodness determination rule learning unit that learns a combination of actual performance data and a control operation of the target plant equipment and a combination of goodness of a control result. A quality determination of a control output is performed in accordance with a combination of actual performance data of a target plant device, control operation, and determination of quality of a control result, and a control rule is learned using the quality determination result, the actual performance data, and supervision data as learning data.

Description

Plant control system, control method, and computer-readable recording medium

Technical Field

The invention relates to a plant control system, a plant control method, and a computer-readable recording medium storing a program.

Background

Conventionally, various types of plant control have been performed based on various control theories in order to obtain an appropriate control result by controlling the plant.

For example, in the rolling mill control, fuzzy control and neuro-fuzzy control are applied as control theories for shape control of the fluctuation state of the control plate. The fuzzy control is applied to shape control using a coolant, and the neuro-fuzzy control is applied to shape control of a sendzimir mill. As shown in patent document 1, shape control using the neuro-fuzzy control is performed to determine a difference between an actual shape pattern (pattern) detected by a shape detector and a target shape pattern and a similarity ratio to a preset reference shape pattern. Then, based on the obtained similarity ratio, a control output amount for the operation end is obtained based on a control rule expressed by controlling the operation amount of the operation end with respect to a preset reference shape pattern.

A conventional technique for controlling the shape of a sendzimir mill using a neuro-fuzzy control will be described below.

In the shape control of the sendzimir mill, the neuro-fuzzy control is used. As shown in fig. 25, the sendzimir mill 50 performs pattern recognition of the shape from the actual shape detected by the shape detector 52 by the pattern recognition unit 51, and calculates which of the reference shape patterns set in advance is the closest to the actual shape. The data of the actual shape detected by the shape detector 52 is subjected to preprocessing for pattern recognition in the shape detection preprocessing section 54.

Then, the control operation unit 53 executes control using a control rule constituted by a control operation end operation amount for a preset shape pattern.

Here, as shown in fig. 26, the pattern recognition unit 51 calculates the actual shape pattern (∈) detected by the shape detector 52 and the target shape (∈)_ref) Is closest to which of the shapes of pattern 1 to pattern 8. Then, the control arithmetic unit 53 selects and executes any one of the control methods of the patterns 1 to 8 based on the arithmetic result.

Documents of the prior art

Patent document

Patent document 1: japanese patent No. 2804161

Patent document 2: japanese patent laid-open publication No. 2018-180799

Disclosure of Invention

Problems to be solved by the invention

In the conventional technique described in patent document 1, a representative shape is set as a reference shape pattern in advance, and control is performed based on a control rule indicating a relationship with a control operation end operation amount with respect to the reference shape pattern. The learning of the control rule is also learning related to the control operation end operation amount with respect to the reference waveform pattern, and a predetermined representative reference shape pattern is used as it is. Therefore, there is a problem of shape control that reacts only to a specific shape pattern.

The reference shape pattern is a pattern that is determined by a person in advance based on knowledge about the target rolling mill and experience of accumulating actual shape results and manual intervention operations, but it is difficult to span all shapes generated in the target rolling mill and the material to be rolled. Therefore, when a shape different from the reference shape pattern is generated, the control by the shape control is not executed, and the shape deviation is not suppressed and remains, or the reference shape pattern is erroneously recognized as being similar, and an erroneous control operation is performed, and the shape may be deteriorated.

Therefore, in the conventional shape control, since the control rule is learned and controlled by using a preset reference shape pattern and a control rule for the reference shape pattern, there is a problem that improvement of the control accuracy is limited.

To solve this problem, for example, a technique described in patent document 2 is proposed. Patent document 2 describes a process of gradually intelligentizing a neural network by learning while causing interference during control. However, the process of generating the control disturbance described in patent document 2 is a process of actually operating the plant device to be controlled, and the process of generating the control disturbance at the time of operating the plant device to be controlled disturbs the actual operation of the plant device to be controlled, and is not preferable in terms of operation. Further, the neural network does not become appropriate unless the plant to be controlled is operated to some extent, and there is a high possibility that appropriate control cannot be performed in a short period from the initial stage of operation.

The purpose of the present invention is to provide a plant control system, a plant control method, and a program that are capable of reducing the risk of disturbing the control of plant and efficiently correcting a control rule.

Means for solving the problems

In order to solve the above problem, for example, the structure described in the claim is adopted.

The present application includes a plurality of means for solving the above-described problems, and by way of example, the plant control system is applied to a system for identifying a pattern of a combination of actual performance data of a plant to be controlled with respect to the plant to be controlled and executing control.

Further, the plant control system includes: a control method learning unit that learns a combination of actual performance data and a control operation of a control target plant device; a control execution unit that executes control of the plant equipment to be controlled based on a combination of the actual performance data learned by the control method learning unit and the control operation; and a goodness determination rule learning unit that learns a combination of actual performance data and a control operation of the control target plant equipment and a combination of goodness of a control result.

Here, the control execution unit includes:

a control rule execution unit that provides a control output in accordance with a combination of actual performance data of the plant equipment to be controlled and the determination of the control operation;

a control output quality determination rule execution unit that performs quality determination of a control output in accordance with a combination of actual performance data of a plant to be controlled, control operation, and determination of quality of a control result;

a new search operation amount calculation unit that calculates an operation amount for searching for a new operation based on the goodness determination in the control output goodness determination rule execution unit; and

and a control output suppression unit that prevents the control output from being output to the plant device to be controlled when it is determined that the performance data of the plant device to be controlled has deteriorated when the control output is output to the plant device to be controlled, using the goodness determination by the control output goodness determination rule execution unit and the simulation data using the control simulator.

In addition, the good/bad decision rule learning unit further includes:

a control result quality determination unit that determines the quality of the control result after a time delay until the control effect is expressed in the actual result data when the control execution unit outputs the control output to the plant to be controlled; and

and a goodness determination rule learning unit that learns the goodness of the control result, the achievement data, and the control output of the control result in the control result goodness determination unit as learning data.

Also, the control method learning unit further includes:

a learning data creation unit that obtains supervisory data using the control output and the quality determination of the control output in the control output quality determination rule execution unit; and

and a control rule learning unit for learning the actual result data and the supervision data as learning data.

Effects of the invention

According to the present invention, it is possible to automatically correct a control rule of a shape pattern and an operation method used for shape control in control with high efficiency while reducing a risk to a plant device, and to obtain an optimal control rule corresponding to a change in environment of the plant device that occurs with time. Therefore, according to the present invention, it is possible to improve the control accuracy, shorten the activation period of the control unit, and cope with a change over time.

Further, according to the present invention, by evaluating the performance of the control rule in advance, there are effects that the risk to the plant equipment is reduced by the application of a new control rule and the control performance is improved by the selection of an optimal control rule.

Problems, structures, and effects other than those described above will become apparent from the following description of the embodiments.

Drawings

Fig. 1 is a schematic configuration diagram of a plant control system according to an example of an embodiment of the present invention.

Fig. 2 is a diagram showing a specific configuration example of the control rule execution unit according to the embodiment of the present invention.

Fig. 3 is a block diagram showing an example of the control output quality determination rule execution unit according to the embodiment of the present invention.

Fig. 4 is a diagram showing a specific configuration example of the control rule learning unit according to the embodiment of the present invention.

Fig. 5 is a configuration diagram showing an example of the merit determination rule learning section according to the embodiment of the present invention.

Fig. 6 is a diagram showing an example of the determination of the quality of the control result of the control method in the shape control of the sendzimir mill.

Fig. 7 is a diagram showing a neural network configuration used for shape control of a sendzimir mill according to an example of the embodiment of the present invention.

Fig. 8 is a diagram illustrating a shape deviation and a control method according to an example of an embodiment of the present invention.

Fig. 9 is a configuration diagram showing an example of a control input data creating unit according to an example of an embodiment of the present invention.

Fig. 10 is a configuration diagram showing an example of a control output calculation unit according to an example of an embodiment of the present invention.

Fig. 11 is a diagram showing a neural network configuration used for the determination of the merits of the sendzimir rolling mill according to the embodiment of the present invention.

Fig. 12 is a diagram showing an operation amount calculation method in the new search operation amount calculation unit according to the example of the embodiment of the present invention.

Fig. 13 is a configuration diagram showing an example of the control output determination unit according to the embodiment of the present invention.

Fig. 14 is a configuration diagram showing an example of a control output calculating unit according to an example of the embodiment of the present invention.

Fig. 15 is a diagram showing the processing stages and the processing contents in the learning data creating unit according to the example of the embodiment of the present invention.

Fig. 16 is a diagram showing an example of data stored in the learning data database according to an example of the embodiment of the present invention.

Fig. 17 is a diagram showing an example of a neural network management table according to an example of the embodiment of the present invention.

Fig. 18 is a configuration diagram showing an example of the learning data database according to an example of the embodiment of the present invention.

Fig. 19 is a configuration diagram showing an example of the control quality determination unit according to the example of the embodiment of the present invention.

Fig. 20 is a diagram showing an example of data stored in the learning data database according to the example of the embodiment of the present invention.

Fig. 21 is a diagram showing an example of a neural network management table according to an example of the embodiment of the present invention.

Fig. 22 is a diagram showing an example of a learning data database according to an example of an embodiment of the present invention.

Fig. 23 is a configuration diagram showing an example in which a plant control system according to an embodiment of the present invention includes a control rule evaluation unit.

Fig. 24 is a block diagram showing an example of a hardware configuration of a plant control system according to an example of the embodiment of the present invention.

Fig. 25 is a structural diagram showing an example of the sendzimir rolling mill.

Fig. 26 is a diagram showing an example of a list of control rules in controlling the shape of the sendzimir mill.

Description of the reference numerals

1 … control target plant equipment, 2 … control input data creation unit, 3 … control output calculation unit, 4 … control output suppression unit, 5 … control output determination unit, 6 … control result goodness determination unit, 7 … learning data creation unit, 10 … control rule execution unit, 16 … control operation disturbance generation unit, 17 … control output goodness determination rule execution unit, 18 … control output operation method selection unit, 20 … control execution unit, 21 … control method learning unit, 22 … goodness determination rule learning unit, 23 … control rule evaluation unit, 31 … goodness determination rule learning unit, 33 … new search operation amount calculation unit, 34 … goodness determination rule accuracy verification unit, 35 … control rule goodness determination data collection unit, 36 … control rule evaluation data calculation unit, 37 … control rule database update unit, 50 … giga rolling mill, 51 … a pattern recognition unit, 52 … a shape detector, 53 … a control calculation unit, 54 … a shape detection preprocessing unit, 101 … a neural network, 102 … a neural network selection unit, 110 … a neural network processing unit, 111 … a neural network, 112 … a neural network learning control unit, 113 … a neural network selection unit, 114 … an input data creation unit, 115 … a supervisory data creation unit, 171 … a neural network, 172 … a neural network selection unit, 201 … a normalized shape deviation, 202 … a shape deviation stage, 210 … a shape deviation PP value calculation unit, 211 … a shape deviation stage calculation unit, 310 … a neural network processing unit, 311 … a neural network, 312 … a neural network learning control unit, 313 … a neural network selection unit, 314 … an input data creation unit, 315 … a supervisory data creation unit, a 501 … a rolling phenomenon model, 502 … a shape correction merit/disadvantage determination unit, 503 a shape deviation prediction data prediction unit, 503 … a shape deviation prediction data prediction correction amount 501, and a rolling phenomenon model, 504 … shape deviation performance data, 505 … shape deviation prediction data, 602 … shape change quality determination unit, 801 … learning data creation unit, 802 … control rule learning unit, DB1 … control rule database, DB2 … learning data database, DB3 … output determination database, DB4 … quality determination database, DB5 … quality determination rule database, DB6 … learning data database, DB7 … verification data database, DB8 … control rule evaluation data database, and DB9 … control rule evaluation value database.

Detailed Description

A plant control system according to an example (hereinafter referred to as "this example") of an embodiment of the present invention will be described below with reference to the drawings.

First, before the present example is described, a case where the plant control system is applied to a shape control device of a rolling mill will be described as an example in which the present invention is completed and its outline.

First, in order to obtain a plant control system capable of efficiently correcting a control rule while reducing a risk to a plant, which is one of the objects of the present invention, the following requirements (1), (2), (3), and (4) are required.

Essential element (1): in order to improve the control rule, if a control operation with a good control result cannot be learned, the control operation is changed greatly, if the control result is good, the control operation is taken in as a new control operation method, if the control operation with a good control result can be learned, the control operation is not changed or is changed only slightly, and if the control result for the situation is good, the control operation is taken in as a new control operation method.

Essential element (2): a combination of a shape pattern, a control operation, and a quality of a control result is learned based on real machine data, thereby constructing a model capable of estimating the quality of the control result with higher accuracy than a simulator using a machine model, and a model suitable for the latest plant equipment state at all times is constructed by automatic learning at regular intervals.

Essential element (3): a model for estimating the quality determination of a control result is used to improve the reliability of a control output suppression function to plant equipment, which is conventionally performed only by a simple mechanical model.

Essential element (4): in a function for generating control rule learning data performed in primary determination of the quality of a control result in the related art, by using a model for estimating the quality of the control result, it is possible to suppress the influence of noise included in plant data, to perform fine adjustment with a small effect as a target of learning data, and to prevent erroneous determination of the control effect, to suppress variation in the learning data, and to stabilize the control performance.

In order to realize these requirements (1) to (4), it is preferable to configure a neural network in the control device, which is capable of learning a combination of the shape pattern used for the shape control, the control operation, and the superiority and inferiority of the control result. Then, the control device needs to estimate the quality of the control result based on the output of the control operation for the shape pattern generated by the rolling mill, using the value obtained by inputting the shape pattern generated by the rolling mill and the output of the control operation to the neural network. The control device selects a method of calculating the control operation amount for searching for a new control operation using the estimated value of the quality of the control result.

The control device does not output to the control operation end of the rolling mill to prevent the shape from deteriorating, for the output which is verified by using a simple model of the rolling mill or the like and the shape is considered to be remarkably deteriorated. In this case, the control device uses the estimated value of the quality of the control result for the determination of the output suppression, thereby improving the reliability of the protection and optimizing the suppression range, thereby making it possible to expand the range of the control function that can be handled.

In the initial stage of application in which the accuracy of estimation of the quality of the control result is insufficient, it is necessary to output the control operation output estimated to be poor to the plant equipment, thereby expanding the learning range for the combination of the shape pattern not learned, the control operation, and the quality of the control result.

At a stage when the accuracy of the quality determination estimation of the control result is sufficiently high, the quality of the control result can be estimated without outputting the operation amount to the plant equipment, and therefore the control device can generate the learning data of the control rule.

The control device estimates the quality of the control result by using a neural network that can estimate the quality of the control result, and can reduce the influence of noise on plant data and determine the quality of fine adjustment data with a small effect. Thereby, the control device can generate learning data. The control device can improve the accuracy of the learning data by preventing erroneous quality determination due to noise.

Further, when the accuracy of estimation of the quality determination of the control result is degraded due to environmental changes of the plant equipment or the like occurring over time, the control device can estimate the quality determination of the control result suitable for the latest plant equipment state by performing relearning using the actual performance data of the nearest plant equipment.

In order to confirm the estimation accuracy of the quality determination of the control result, test data for accuracy verification is prepared separately from data used for learning of the neural network. Then, the control device can confirm the estimation accuracy of the goodness determination based on an error of the goodness determination of the value output by inputting the shape pattern included in the test data for accuracy verification and the output of the control operation to the neural network and the control result included in the test data.

Fig. 1 shows a configuration of a plant control system of this example.

The plant control system of fig. 1 includes a control execution unit 20, a control method learning unit 21, a quality determination rule learning unit 22, a plurality of databases DB (DB1 to DB8), and a management table TB for each database DB.

The control execution means 20 inputs the actual performance data Si from the control target plant 1, and supplies the control operation amount output SO determined in accordance with the control rule (neural network) to the control target plant 1 to control the control target plant 1. The controlled plant facility 1 is the sendzimir mill 50 shown in fig. 25, which has been described above.

Here, the control rule is, for example, a control rule for a shape pattern a (∈) of a detected actual result and a target shape (∈) as described with reference to fig. 26_ref) The difference a (Δ ∈) of (b) is closest to a rule of which one of the shapes of the prepared plurality of patterns is calculated. The control execution unit 20 selects and executes a control method of any pattern based on the operation result of the control rule.

The control method learning means 21 inputs the control input data S1 and the like created by the control execution means 20, performs learning, and reflects the learned control rule in the control execution means 20.

The merit judgment rule learning means 22 inputs the performance data Si before and after the control of the plant 1 to be controlled, and the like, performs learning, and reflects the learned merit judgment rule in the control execution means 20.

The control execution means 20 includes a control input data creation unit 2, a control rule execution unit 10, a control output calculation unit 3, a control output suppression unit 4, a control output determination unit 5, a control output merit determination rule execution unit 17, a new search operation amount calculation unit 33, and a control output operation method selection unit 18.

The control execution means 20 uses the control input data creation unit 2 to create the input data S1 of the control rule execution unit 10 based on the actual result data Si of the rolling mill as the control target plant 1.

The control rule execution unit 10 performs a control rule execution process of generating a control operation terminal operation command S2 from the actual result data Si of the control target, using a neural network (control rule) expressing the relationship between the actual result data Si of the control target and the control operation terminal operation command S2. The control output arithmetic unit 3 calculates a control operation amount S3 for the control operation side based on the control operation side operation command S2. Thus, the control execution unit 20 generates the control manipulated variable S3 using the neural network based on the actual result data Si of the control target plant 1.

The control output goodness determination rule execution unit 17 performs control output goodness determination rule execution processing for generating a control output goodness determination estimate S9 from the control target goodness data Si and the control manipulated variable S3, using a neural network (goodness determination rule) expressing the relationship between the control target goodness data Si, the control manipulated variable S3, and the control result goodness data S6 of the control result. The control output goodness determination rule execution unit 17 generates a control result goodness determination estimated value S11 from the control target performance data Si and the selection control operation amount S8 described later.

The new search manipulated variable arithmetic unit 33 performs new search manipulated variable arithmetic processing for calculating a new search control manipulated variable S12 based on the control output goodness determination estimated value S9.

The control output operation method selection unit 18 creates a selection control operation amount S8 and a control method selection flag S14 based on the control operation amount S3 or the new search control operation amount S12.

The control output determination unit 5 in the control execution unit 20 determines the control operation amount output availability data S4 for the control operation end, using the performance data Si from the control target plant 1 and the control operation amount S3 from the control output calculation unit 3. The control output suppressing unit 4 determines whether or not the selected control manipulated variable S8 can be output to the control manipulating end based on the control manipulated variable output availability data S4 and the control result quality determination estimated value S11, and outputs the selected control manipulated variable S8 as the control manipulated variable output SO to be supplied to the plant 1 to be controlled. Thus, the selected control operation amount S8 determined to be abnormal is not output from the control execution unit 20 to the control target plant 1.

The control execution unit 20 configured as described above refers to the control rule database DB1, the output determination database DB3, and the good/bad determination rule database DB5 in order to execute this processing.

The control rule database DB1 is connected to both the control rule execution unit 10 in the control execution unit 20 and the control rule learning unit 802 in the control method learning unit 21, which will be described later, so as to be accessible.

The control rule (neural network) as a result of learning in the control rule learning section 802 is stored in the control rule database DB 1. The control rule execution part 10 refers to the control rules stored in the control rule database DB 1.

The learning data obtained by the control rule learning unit 802 is stored in the learning data database DB 2.

The output determination database DB3 is connected to the control output determination unit 5 in the control execution unit 20 so as to be accessible, and the output determination result is stored in the output determination database DB 3.

The data used for the determination of superiority is stored in the superiority determination database DB 4.

The goodness determination rule (neural network) as the learning result in the goodness determination rule learning section 31 is stored in the goodness determination rule database DB 5. The quality determination rule database DB5 is connected to both the control output quality determination rule execution unit 17 in the control execution means 20 and the quality determination rule learning unit 31 in the quality determination rule learning means 22, which will be described later, so as to be accessible. The control output good/bad decision rule execution part 17 refers to the good/bad decision rule stored in the good/bad decision rule database DB 5.

The learning data learned by the control method learning unit 21 is stored in the learning data database DB 6.

The verification data DB7 stores therein verification data necessary for making a quality determination.

Fig. 2 shows a specific configuration example of the control rule execution unit 10 of the present example.

The input data S1 created by the control input data creation unit 2 is input to the control rule execution unit 10. The control rule execution unit 10 processes the input data S1 and provides a control operation end operation command S2 to the control output operation unit 3. The control rule execution unit 10 includes a neural network 101, and the neural network 101 outputs a control operation end operation command S2 conforming to the shape control rule as shown in fig. 26.

The control rule execution unit 10 further includes a neural network selection unit 102 that refers to the control rules stored in the control rule database DB1 to select an optimal control rule as a control rule in the neural network 101 and causes the neural network 101 to execute the selected control rule.

In this way, the control rule execution unit 10 selects a neural network necessary for use from a plurality of neural networks divided for each operator group and control purpose. The control rule database DB1 may include performance data (data of operation groups, etc.) Si that enables selection of a neural network and a criterion for determination of superiority or inferiority as data from the plant 1 to be controlled.

Since the execution of the neural network is a control rule, the neural network and the control rule are used synonymously in this specification.

Fig. 3 shows a specific configuration of the control output quality determination rule execution unit 17.

The input data S1 created by the control input data creation unit 2 and the control manipulated variable S3 created by the control output calculation unit 3 are input to the control output merit/disadvantage determination rule execution unit 17. The control output goodness determination rule execution unit 17 generates a control output goodness determination estimation value S9 based on these input data and supplies it to the new search operation amount calculation unit 33.

The input data S1 created by the control input data creation unit 2 and the selection control operation amount S8 created by the control output operation method selection unit 18 are input to the control output quality determination rule execution unit 17. The control output goodness determination rule execution unit 17 generates the control result goodness determination estimation value S11 based on these input data, and supplies it to the control output suppression unit 4.

The control output quality determination rule execution unit 17 includes a neural network 171 and a neural network selection unit 172.

The neural network 171 estimates, based on the past control results, the goodness determination value of the control result in the case where the control manipulated variable S3 (control pattern) is output, with respect to the input data S1 (shape pattern).

The neural network selecting section 172 selects an optimum goodness determination rule as the goodness determination rule in the neural network 171 by referring to the goodness determination rule stored in the goodness determination rule database DB 5.

In this way, the control output quality determination rule execution unit 17 selects a desired neural network from among a plurality of neural networks divided according to the difference in the material properties to be controlled and the difference in the quality determination criterion.

The quality determination rule database DB5 may include performance data (data of operation groups, etc.) Si that enables selection of material properties to be controlled and criteria for quality determination as data from the plant 1 to be controlled. Since the neural network becomes a good/bad judgment rule when executed, the neural network and the good/bad judgment rule are used synonymously in this specification.

Returning to the explanation of fig. 1, the control method learning unit 21 performs learning of the neural network 101 used in the control execution unit 20.

The control method learning unit 21 includes a learning data creation unit 801 and a control rule learning unit 802.

The learning data creation unit 801 in the control method learning unit 21 performs a learning data creation process of creating new supervision data S7a used for learning the neural network, using the selection control operation amount S8 created by the control execution unit 20, the control method selection flag S14, and the control result goodness determination estimated value S11 created by the control output goodness determination rule execution unit 17. The supervision data S7a created by the learning data creation unit 801 is supplied to the control rule learning unit 802.

The supervisory data S7a corresponds to the control operation side operation command S2 output by the control rule execution section 10.

The learning data creation unit 801 obtains, as new supervision data S7a, data obtained by estimating the control operation terminal operation command S2 output by the control rule execution unit 10, using the control result goodness determination estimation value S11 created by the control output goodness determination rule execution unit 17.

Fig. 4 shows a specific configuration example of the control rule learning unit 802.

The control rule learning unit 802 includes an input data creation unit 114, a supervisory data creation unit 115, a neural network processing unit 110, and a neural network selection unit 113.

The control rule learning unit 802 receives input data S1 from the control input data creation unit 2 and new supervisory data S7a from the learning data creation unit 801 as inputs from the outside. The control rule learning unit 802 refers to data accumulated in the control rule database DB1 and the learning data database DB 2.

In the control rule learning unit 802, the input data S1 is taken into the neural network processing unit 110 via the input data creation unit 114.

In the control rule learning unit 802, the new supervision data S7a from the learning data creation unit 801 is supplied to the neural network processing unit 110 as the total supervision data S7c including the past supervision data S7b stored in the learning data database DB2 in the supervision data creation unit 115. These pieces of supervision data S7a and S7b are stored in the learning data database DB2 as appropriate and used.

Similarly, the input data S8a from the control input data creating unit 2 is supplied to the neural network processing unit 110 in the form of the total input data S8c including the past input data S8b stored in the learning data database DB2 in the input data creating unit 114. These input data S8a and S8b are also stored in the learning data database DB2 and used as appropriate.

The neural network processing unit 110 includes a neural network 111 and a neural network learning control unit 112.

The neural network 111 takes in the input data S8c from the input data creating unit 114, the supervisory data S7c from the supervisory data creating unit 115, and the control rule (neural network) selected by the neural network selecting unit 113, and stores the neural network that is finally determined in the control rule database DB 1.

The neural network learning control unit 112 controls the input data creation unit 114, the supervisory data creation unit 115, and the neural network selection unit 113 at appropriate times to obtain inputs to the neural network 111, and stores the processing results in the control rule database DB 1.

Here, the neural network 101 in the control rule execution unit 10 of fig. 2 and the neural network 111 in the control method learning unit 21 of fig. 4 are both neural networks having the same concept, but there is a difference as described below.

That is, the neural network 101 in the control rule execution unit 10 is a predetermined neural network, and when the input data S1 is supplied, it obtains the control operation terminal operation command S2 as a corresponding output.

On the other hand, the neural network 111 in the control method learning unit 21 is a neural network that finds the input-output relationship by learning when the input data S1 and the input data S8c and the supervisory data S7c of the control operation terminal operation command S2 are set as learning data.

A method of consideration of the basic processing in the control method learning unit 21 is as follows.

First, when the content of the control operation amount output availability data S4 is "available", the control execution unit 20 outputs the control operation amount output SO to the control target plant 1. Here, when the content of the control result goodness determination estimation value S11 is "good" (i.e., changes in the direction toward the improvement of the achievement data Si), the learning data creation unit 801 determines that the selection control operation amount S8 output by the control output operation method selection unit 18 is correct, and creates learning data so that the output of the neural network becomes the selection control operation amount S8.

On the other hand, when the content of the control operation amount output availability data S4 is "no" or the content of the control result goodness determination estimate value S11, which is the control operation amount output SO output to the plant 1 to be controlled, is "no" (a change in the direction in which the performance data Si becomes worse), the learning data creation unit 801 determines that the selected control operation amount S8 output by the control output operation method selection unit 18 is erroneous.

In this case, the learning data creation unit 801 checks whether or not the control operation amount S3 is selected by the control output operation method selection unit 18, based on the control method selection flag S14. When the control manipulated variable S3 is selected in this confirmation, the learning data creation unit 801 determines that the control manipulation end manipulation command S2 output by the control rule execution unit 10 is erroneous, and creates learning data so that the output of the neural network is not output. At this time, the neural network output is configured to output 2 kinds of outputs of + direction and-direction to the same control operation terminal as the control output, and the learning data is created so that the control operation terminal operation command S2 on the output side is not output.

The control rule learning unit 802 shown in fig. 4 performs the following processing as a result of data processing by the neural network learning control unit 112.

First, the control rule learning unit 802 performs learning of the neural network 101 used by the control rule execution unit 10, using learning data that is a combination of the data S8c obtained from the input data S1 for the control execution means 20 and the supervision data S7c created by the supervision data creation unit 115.

Actually, the control rule learning unit 802 includes the same neural network 111 as the neural network 101 of the control rule execution unit 10 in the control rule learning unit 802, and performs an operation test under various conditions to learn a response at that time, and obtains a control rule that is confirmed to produce a better result as a result of the learning.

Since the learning needs to be performed using a plurality of learning data, a plurality of learning data in the past are taken out from the learning data database DB2 in which learning data generated in the past are accumulated, and the learning is performed to execute the processing. Then, the learning data of this time is stored in the learning data database DB 2. The learned neural network is stored in the control rule database DB1 for use in the control rule executing unit 10.

The learning of the neural network may be performed by using the past learning data together each time new learning data is created, or may be performed by using the past learning data together after the learning data is accumulated to a certain extent (for example, 100 pieces).

With such a configuration, the control output operation method selection unit 18 selects the new search operation amount, thereby outputting the new search operation amount to the target plant, creating learning data based on the control result, and learning the new control method.

Returning to the explanation of fig. 1, the goodness determination rule learning unit 22 executes the nerve used in the control execution unit 20Learning of network 171 (fig. 3). When the control execution means 20 outputs the control manipulated variable output SO to the plant device 1 to be controlled, the control effect actually appears as a change in the actual result data Si in a long time. Therefore, learning is performed using data that is time-delayed by that amount of time. In FIG. 1 and the like, the term "Z" is used^-1The processing unit DL of "indicates that there is an appropriate time delay in transmitting each data.

The quality determination rule learning means 22 includes a control result quality determination unit 6, a quality determination rule learning unit 31, and a quality determination database DB 4.

The control result goodness determination unit 6 performs a control result goodness determination process for determining whether the performance data Si changes in the direction of making good or making bad, using the performance data Si and the previous value Si0 of the performance data from the control target plant 1 and the goodness determination data S5 stored in the goodness determination database DB 4. Then, the control result quality determination unit 6 outputs control result quality data S6 indicating the determination result.

Fig. 5 shows a specific configuration of the quality determination rule learning unit 31.

The quality determination rule learning unit 31 includes an input data creation unit 314, a supervisory data creation unit 315, a neural network processing unit 310, and a neural network selection unit 313.

The quality determination rule learning unit 31 obtains, as inputs from the outside, data S12a1 obtained by time-delaying the input data S1 from the control input data creation unit 2, and data S12a2 obtained by time-delaying the control output amount S0 from the control output suppression unit 4. The good/bad decision rule learning unit 31 obtains the control result good/bad data S6 from the control result good/bad decision unit 6(S13 a).

The good/bad decision rule learning unit 31 refers to data accumulated in the good/bad decision rule database DB5 and the learning data database DB 6.

The input data S1 and the control output S0 are acquired into the neural network processing unit 310 through the input data creation unit 314 after appropriate time delay compensation.

The control result quality data S6(S13a) from the control result quality determination unit 6 is supplied to the neural network processing unit 310 as the total supervised data S13c including the past supervised data S13b stored in the learning data database DB6 in the supervised data creation unit 315. These pieces of supervision data S13a and S13b are stored in the learning data database DB6 as appropriate and used.

Similarly, the input data S12a1 and S12a2 from the control input data creating unit 2 and the control output suppressing unit 4 are supplied to the neural network processing unit 310 in the form of the total input data S12c including the past input data S12b stored in the learning data database DB6 in the input data creating unit 314. These input data S12a1, S12a2, and S12b are stored in the learning data database DB6 and used as appropriate.

The neural network processing unit 310 includes a neural network 311 and a neural network learning control unit 312.

The neural network 311 takes in the input data S12c from the input data creating unit 314, the supervisory data S13c from the supervisory data creating unit 315, and the control rule (neural network) selected by the neural network selecting unit 313. Then, the neural network 311 stores the finally decided neural network in the goodness determination rule database DB 5.

The neural network learning control unit 312 controls the input data generation unit 314, the supervisory data generation unit 315, and the neural network selection unit 313 at appropriate times to obtain inputs to the neural network 311. The neural network learning control unit 312 stores the processing result in the quality determination rule database DB5 via the neural network selection unit 313.

Fig. 6 is a diagram showing a specific example of the determination of the quality of the control result of the control method in the shape control of the sendzimir mill. Fig. 6 shows the result of the determination of the quality of the control result for each of the shape control rules shown in fig. 26.

Here, the neural network 171 of the control execution unit 20 shown in fig. 3 and the neural network 311 of the merit determination rule learning unit 22 shown in fig. 5 are both neural networks of the same concept, but differ in the following points.

The neural network 171 in the control execution unit 20 is a neural network of a predetermined content. That is, the neural network 171 is a so-called neural network used for processing in one direction, which obtains the control output goodness determination estimated value S9 or S11 as a corresponding output when the input data S1 and the selection control operation amount S8 or the control operation amount S3 are supplied.

In contrast, the neural network 311 in the merit determination rule learning unit 22 is a neural network that satisfies the input-output relationship between the input data S1 and the input data S12c and the supervision data S13c of the control output quantity S0 when they are set as learning data.

Next, a specific example of a plant control method will be described for shape control in the sendzimir mill. The following specification A, B is used to describe the shape control.

The specification a is a specification regarding priority, and has information on priority in the board width direction. For example, in the shape control, it is often difficult to control the mechanical characteristics to a target value over the entire region in the plate width direction. Therefore, specifications a1, a2 regarding the following 2 priorities are set in the board width direction. Here, the specification a1 regarding the priority is a specification of "giving priority to the board end". The priority specification a2 is a specification of "giving priority to the center portion".

Control is performed in accordance with the 2 priorities of the specifications a1 and a 2. That is, in the case where the plant control system performs control, either of the specifications a1 or a2 regarding the priority is considered.

The specification B is a specification for coping with a condition determined in advance. For example, since the relationship between the shape pattern and the control method changes under various conditions, it is necessary to divide the shape pattern into a range of, for example, B1 as a sheet width and B2 as a steel type. The degree of influence on the shape of the shape manipulation end varies by each specification change.

The plant equipment 1 to be controlled in this example is a sendzimir mill, and the actual result data is the shape actual result. The sendzimir mill is a mill having multiple rolls (cluster rolls) for cold rolling of a hard material such as stainless steel. The sendzimir mill uses small diameter work rolls for the purpose of applying high pressure to hard materials. Therefore, it is difficult to control the sendzimir mill to obtain a flat steel sheet. As a countermeasure, the sendzimir mill employs a multi-roll structure and various shape control units.

The sendzimir mill generally includes a single conical first upper and lower intermediate rolls, which are displaceable, and 6 split rolls (split rolls) and 2 rolls called AS-U, which are provided above and below the first intermediate roll. In the example described below, the actual shape data Si is detected by a shape detector, and the input data S1 is a shape deviation that is a difference from the target shape. The control manipulated variable S3 is the AS-U of #1 to # n and the roller shift amounts of the upper and lower first intermediate rollers.

Fig. 7 shows a neural network structure used for shape control of the sendzimir mill. Here, the neural network 101 is shown as a neural network for controlling the rule executing unit 10. The control rule learning unit 802 also represents the neural network 111. The neural network 101 and the neural network 111 are both constructed identically.

In the case of the shape control of the sendzimir mill, the performance data Si from the plant equipment 1 to be controlled is the performance data of the sendzimir mill including the data of the shape detector (here, the shape deviation which is the difference between the output performance shape and the target shape). The control input data creation unit 2 obtains the normalized shape deviation 201 and the shape deviation stage 202 as input data S1. Thus, the input layers of the

neural networks

101, 111 are composed of normalized shape deviations 201 and shape deviation stages 202. In fig. 7, the shape deviation stage 202 is input to the neural network input layer, but the neural network may be switched in accordance with the stage.

The output layer of the

neural networks

101 and 111 corresponds to the AS-U, which is the shape control operation end of the sendzimir mill, and the first intermediate roll, and is composed of an AS-U operation degree 301 and a first intermediate operation degree 302. Regarding each degree of operation, AS for the AS-U, each AS-U has an AS-U opening direction (direction in which a roll gap (interval between upper and lower work rolls of a rolling mill) is opened), and an AS-U closing direction (direction in which the roll gap is closed).

The upper and lower first intermediate rolls have a first intermediate roll opening direction (direction in which the first intermediate roll moves outward from the center of the rolling mill) and a first intermediate roll closing direction (direction in which the first intermediate roll moves toward the center of the rolling mill).

For example, when the shape detector has 20 regions and the shape deviation stage 202 has 3 stages (large, medium, and small), the input layer has 23 inputs. When the number of saddles of the AS-U is 7 and the upper and lower first intermediate rollers can be displaced in the sheet width direction, the AS-U operation degrees 301 for the carry-out layer become 14, and the total number of the first intermediate operation degrees becomes 18, that is, 4. The number of layers of the intermediate layer and the number of neurons in each layer are set in time.

Further, the shape control operation terminal of the sendzimir mill as the output layer constitutes a neural network output so that 2 kinds of outputs of + direction and-direction are output to the respective control operation terminals.

Fig. 8 shows a shape deviation and a control method in this example.

Fig. 8 (a) shows a control method in the case where the shape deviation is large, and fig. 8 (b) shows a control method in the case where the shape deviation is small. In fig. 8, (a) and (b) show the shape deviation in the height direction (vertical axis direction), the width direction in the horizontal axis direction, the plate end portions at both ends of the width, and the plate center portion at the center.

As shown in fig. 8 (a), when the shape deviation is large, the entire shape is corrected with priority over the local shape deviation in the plate width direction.

On the other hand, as shown in fig. 8 (b), when the shape deviation is small, it is prioritized to reduce the local shape deviation.

Since it is necessary to change the control method according to the magnitude of the shape deviation, the shape deviation stage 202 is provided as shown in fig. 7 and is supplied to the

neural networks

101 and 111, and the magnitude of the shape deviation is determined. The shape deviation can be normalized to 0 to 1, for example, regardless of the size of the shape deviation. This is an example, and the shape deviation may be input directly to the input layer of the neural network without being normalized, or the neural network itself may be changed according to the magnitude of the shape deviation. For example, 2 neural networks may be prepared and divided into a neural network used when the shape deviation is large and a neural network used when the shape deviation is small.

In the plant control of this example, the operation method for the shape pattern is learned for the

neural networks

101 and 111 having the configuration as described above with reference to fig. 7, and the shape control is executed using the learned neural network. Even in the case of neural networks having the same structure, the characteristics are different depending on the learning conditions, and different control outputs can be output for the same shape pattern.

Therefore, by separately using a plurality of neural networks according to other conditions of the shape performance, it is possible to configure optimal control for a plurality of conditions. This is a countermeasure for the specification B. The configuration of fig. 2 described above shows a specific example in the case where such a specification is performed.

That is, in the configuration example of fig. 2, different neural networks are prepared from the rolling result, the name of the rolling mill operator, the type of steel of the material to be rolled, the plate width, and the like, for the neural network 101 used in the control rule execution unit 10, and are registered in the control rule database DB 1. The neural network selecting unit 102 selects a neural network that matches the condition at the time point, and sets the neural network as the neural network 101 of the control rule executing unit 10.

As a condition at this point in time in the neural network selecting unit 102, data of the plate width may be acquired from the actual performance data Si in the plant 1 to be controlled, and the neural network may be selected in accordance with the data. As long as the plurality of neural networks used here have input layers and output layers as shown in fig. 7, the number of intermediate layers and the number of cells in each layer may be different.

Fig. 9 shows a configuration of the control input data creation unit 2 for creating data S1 (normalized shape deviation 201, shape deviation stage 202) to be input to the input layers of the

neural networks

101 and 111.

The control input data creation unit 2 receives shape detector data of a shape detector for detecting a plate shape during rolling in a sendzimir mill as the plant 1 to be controlled as actual result data Si. Then, the control input data creating unit 2 obtains a shape deviation PP value (Peak To Peak value) SPP, which is a difference between the maximum value and the minimum value of the detection results of the respective shape detector regions, by the shape deviation PP value calculating unit 210.

The shape deviation stage calculation unit 211 classifies the shape deviation into 3 stages of large, medium, and small based on the shape deviation PP value SPP. The shape is a distribution of the elongation of the rolled material in the width direction of the sheet, and I to UNIT each representing the elongation by 10-5 UNITs is used as a UNIT. For example, classification is performed as shown in the following numerical expression.

Here, the shape deviation stages are classified into (large-to-1, medium-to-0, and small-to-0) by the establishment of [ expression 1], into (large-to-0, medium-to-1, and small-to-0) by the establishment of [ expression 2], and into (large-to-0, medium-to-0, and small-to-1) by the establishment of [ expression 3 ]. Regarding the shape deviation of each region, the normalization is performed using SPM, which is referred to as SPM ═ SPP.

[ number formula 1]

SPP≥50I-UNIT

[ numerical formula 2]

50I-UNIT＞Spp≥10I-UNIT

[ numerical formula 3]

As described above, the control input data creation UNIT 2 creates the normalized shape deviation 201 and the shape deviation stage 202, which are input data to the neural network 101, 10I-UNIT > Spp. The normalized shape deviation 201 and the shape deviation stage 202 are input data S1 of the control rule execution unit 10.

Fig. 10 shows a configuration of the control output arithmetic unit 3.

The control output arithmetic unit 3 generates a control operation amount S3 as an operation command to each shape control operation side, based on the control operation side operation command S2 as an output from the neural network 101 in the control rule execution unit 10. The control operation end operation command S2 corresponds to the AS-U operation degree 301 and the first intermediate operation degree 302 in the shape control of the sendzimir mill.

Fig. 10 shows 1 data example of each of a plurality of AS-U operation degrees 301 and a plurality of first intermediate operation degrees 302, each data being composed of a pair of data of an opening direction degree and a closing direction degree.

In the control output arithmetic section 3, the inputted AS-U operation degree 301 has outputs in the AS-U opening direction and the AS-U closing direction, and therefore the difference between them is calculated by the subtractor 303. Then, the output of the subtractor 303 is multiplied by a conversion gain G in a multiplier 304_ASUAnd generating and outputting an operation instruction for each AS-U. Since the control output to each AS-U is the AS-U position change amount (unit is length), the gain G is converted_ASUThe gain is converted from the operation degree to the position change amount.

The first intermediate degree of operation 302, which is likewise input, has a first intermediate outer, inner output, so their difference is calculated by subtractor 305. Then, the output of the subtractor 305 is multiplied by a conversion gain G by a multiplier 306_1STThereby generating and outputting an operation instruction for each first intermediate roller shift. Since the control output to each first intermediate roller becomes the first intermediate roller shift position change amount (unit is length), the gain G is converted_1STThe gain is converted from the operation degree to the position change amount.

According to the above method, the control output calculation unit 3 can calculate the control manipulated variable S3. The control manipulated variable S3 includes #1 to # nAS-U position change amounts (n depends on the number of saddles of the AS-U roller), an upper first intermediate shift position change amount, and a lower first intermediate shift position change amount.

Fig. 11 shows a neural network configuration for controlling the goodness determination for outputting the shape control result of the sendzimir mill used in the goodness determination rule execution unit 17 and the goodness determination rule learning unit 31. The neural network here is represented as the neural network 171 when used for controlling the output of the merit determination rule executing section 17, and the neural network 311 when used for the merit determination rule learning section 31, but the configurations are the same.

As the input data S1, the same signals as those inputted to the neural network input layer described in fig. 7 are used for the normalized shape deviation 201 and the shape deviation stage 202. Then, the control manipulated variable S3 or a selection control manipulated variable S8 described later is input to the input layer. The control operation amount S3 or the selection control operation amount S8 is constituted by a position change amount of each control operation device.

The output layer outputs a value for determining whether the estimated control result is good or bad when the control manipulated variable S3 or the selection control manipulated variable S8 is output to the input data S1. The number of layers of the intermediate layer and the number of neurons in each layer are set in time.

Fig. 12 shows an operation amount calculation method in the new search operation amount calculation unit 33.

The new search manipulated variable calculation unit 33 calculates a new search control manipulated variable S12 using the control output goodness determination estimate S9 output from the control output goodness determination rule execution unit 17 in accordance with the following policy.

That is, since the goodness determination of the control operation is estimated favorably when the value of the control output goodness determination estimated value S9 is large, the new search operation amount calculation unit 33 performs fine adjustment as the new search operation amount.

When the value of the control output quality determination estimated value S9 is small, the quality determination of the control operation is estimated to be poor, and therefore the new search operation amount calculation unit 33 searches for a new appropriate operation method by largely changing the control operation.

Based on the above-described policy, an expression for obtaining the new search operation amount Crand is set as follows.

IF(S9>th)THEN Crand＝Cref^*(1+β^*th1)

IF(th≥S9)THEN Crand＝Cref+γ^*th2^*G

Here, β and γ represent random values generated between-1 and 1. th1 indicates the degree of fine adjustment to be made,

for example, when the fine adjustment is performed in a range of ± 10% of the original command, th1 is set to 0.1.

th2 is a setting for largely changing the degree of the operation method, and for example, when th2 is set to 0.1, an offset of 10% is added to the original command, and there is a possibility that the operation polarity is changed or a command of a device which is not originally operated is output.

The values of β and γ are different for each operation device, and the operation amount of each device is independently changed. G denotes a maximum operation position control command for each control operation device, and the value of% is converted into an operation position control command by multiplying the command% described above.

The control output operation method selection portion 18 selects the control operation amount S3 or the new search control operation amount S12 and outputs it as the selected control operation amount S8. The selection control operation amount S3 and the new search control operation amount S12 are determined probabilistically, and the probability Prand of using the new search control operation amount S12 can be set to 0 to 1 by the user. The value δ is determined by the following equation, using a value that is randomly set to 0 to 1.

IF(δ>Prand)THEN C”ref＝Cref、α＝1

ELSE C”ref＝Crand、α＝0

Here, C "ref represents the selection control operation amount S8 that the control output operation method selection unit 18 outputs to the subsequent operation units. δ uses a shared value in the calculation of the operation amounts of all the devices, and all the devices use the operation amount of the same side. α is a control method selection flag S14, and takes 1 when the control operation amount S3 is selected and takes 0 when the new search control operation amount S12 is selected. The control method selection flag S14 is output to the subsequent arithmetic unit together with the selection control operation amount S8. As a method of setting Prand, in the actual machine control, 0 is set when it is not desired to give a risk to the plant equipment by random operation, and a ratio other than 0 is set when it is desired to output a new operation amount for search in order to improve the control rule.

Fig. 13 shows a configuration of the control output determination unit 5.

The control output determination unit 5 includes a rolling phenomenon model 501 and a shape correction quality determination unit 502. The control output determination unit 5 obtains the performance data Si from the plant 1 to be controlled, the control manipulated variable S3 from the control output calculation unit 3, and the information of the output determination database DB3, and provides the control manipulated variable output availability data S4 to the control operation end.

The control output determination unit 5 having such a configuration predicts a change in shape when the selected control operation amount S8 calculated by the control output operation method selection unit 18 is output to the rolling mill as the control target plant 1 by inputting the change to the model of the known control target plant 1. The model of the known plant 1 to be controlled is here a rolling phenomenon model 501. In this prediction, when the shape is expected to deteriorate, the control output determination unit 5 suppresses the control manipulated variable output SO to prevent the shape from greatly deteriorating.

More specifically, the control output determination unit 5 inputs the selection control operation amount S8 to the rolling phenomenon model 501, predicts the shape change caused by the selection control operation amount S8, and calculates the shape deviation correction amount prediction data 503.

On the other hand, the control output determination unit 5 adds the shape deviation correction amount prediction data 503 to the shape detector data Si from the control target plant 1 to obtain shape deviation prediction data 505, and evaluates the shape deviation prediction data 505. Thus, the control output determination unit 5 can predict how the shape changes when the control manipulated variable S3 is output to the plant 1 to be controlled. The shape detector data Si here is the shape deviation performance data 504 at the current time point.

The control output determination unit 5 determines whether the shape correction performance determination unit 502 changes in the direction of the shape becoming better or in the direction of the shape becoming worse based on the current shape deviation performance data 504 and the shape deviation prediction data 505, and obtains control operation amount output availability data S4.

Specifically, the shape correction quality determination unit 502 performs quality determination of the shape correction as follows. First, as shown in specification A, B regarding the priority of shape control, in order to consider the control priority in the board width direction, a weight coefficient w (i) in the board width direction is set in advance for each of specification a1 and specification a2 in output determination database DB 3. Using the weight coefficient, the quality of the shape change is determined using, for example, an evaluation function J such as the following expression 4. In equation 4, w (i) is a weight coefficient, ∈ fb (i) is shape deviation actual result data 504, ∈ est (i) is shape deviation prediction data 505, i is a shape detector area, and rand is a random number term.

[ numerical formula 4]

When the evaluation function J of [ equation 4] is used, the evaluation function J becomes positive when the shape is good, and becomes negative when the shape is bad. The rand is a random number term, and changes the evaluation result of the evaluation function J randomly. Thus, even when the shape is deteriorated, the evaluation function J is positive, and therefore, even when the rolling phenomenon model 501 is not correct, the relationship between the shape pattern and the control method can be learned.

Here, the random number term rand is changed in time as follows: as in the first trial operation, the maximum value is increased when the model of the plant 1 to be controlled is not reliable, and 0 is set when it is desired to learn the control method to some extent and execute stable control.

The shape correction quality determining unit 502 calculates the evaluation function J, outputs the control manipulated variable output availability data S4 as control manipulated variable output availability data S4 as 0 (no) when J is equal to or greater than 0 and outputs the control manipulated variable output availability data S4 as 1 (available) when J is equal to or greater than 0.

As described above, the normalized shape deviation 201, the shape deviation stage 202, and the selection control manipulated variable S8 are input to the control output quality determination rule execution unit 17, and the control result quality determination estimated value S11 is output. The control result goodness determination estimation value S11 takes a value of 1 when it is estimated that the result of control is good, and takes a value of 0 in addition.

The control output suppression unit 4 determines whether or not to output the control manipulated variable output SO to the plant 1 to be controlled, based on the control manipulated variable output availability data S4 and the control result goodness determination estimate value S11, which are the determination results of the control output determination unit 5. The control manipulated variable output availability data S4 is a #1 to a # nAS-U position change amount output, an upper first intermediate shift position change amount output, and a lower first intermediate shift position change amount output, and is determined by the following conditions.

IF (control method selection flag ═ 1) THEN

IF (control manipulated variable output availability data S4 being 0OR control result goodness determination estimated value S11 ≦ thprot) THEN

Position change amount output of #1 to # nAS-U is 0

Upper first intermediate shift position change amount output equals 0

The next first intermediate shift position change amount output is equal to 0

ELSE

The output of the position change amounts #1 to # nAS-U is the position change amounts #1 to # nAS-U

The upper first intermediate shift position change amount output is equal to the upper first intermediate shift position change amount

The lower first intermediate shift position change amount output is equal to the lower first intermediate shift position change amount

ENDIF

ELSE

IF (0 OR control result goodness determination estimate S11 ≦ thprot) AND (PTRIAL < η) THEN (control manipulated variable output availability data S4 ≦ thren)

Position change amount output of #1 to # nAS-U is 0

Upper first intermediate shift position change amount output equals 0

The next first intermediate shift position change amount output is equal to 0

ELSE

ENDIF

Here, thprot sets a reference value for suppressing the output, based on the estimated value for determining the quality of the control result. Specifically, it is considered that the estimation accuracy of the goodness determination is low even at the initial stage of startup when the operation data of the plant is insufficient, and therefore the reference value is lowered in advance, and the output suppression based on the estimation of the goodness determination is not performed so much.

On the other hand, after the actual result data of the operation is sufficiently accumulated and the accuracy of the goodness determination is increased, the reference value is increased, and the effect of suppressing the output estimated based on the goodness determination of the control result is improved. Regarding the accuracy of the goodness determination, the determination is made based on the result of the verification of the estimation accuracy of the goodness determination rule currently used, by receiving the goodness determination rule accuracy S15 from the goodness determination rule accuracy verification portion 34 in the goodness determination rule learning unit.

Further, η is a variable having a random value in the range of 0 to 1, and ptial represents a probability that the output suppression is invalidated and a new search operation is output to the plant. When the control method selection flag S14 is 0, the effect of the control method in the unknown area is verified, and therefore, the output suppression to the plant equipment is ignored with a certain probability and the plant equipment is output.

The control execution means 20 executes the above-described calculation based on the actual result data Si from the plant 1 (rolling mill) to be controlled, and outputs the control manipulated variable output SO to the plant 1 (rolling mill) to be controlled, thereby executing the shape control. In addition, the control method learning unit 21 uses data used in the control execution unit 20.

Next, the operation performed by the learning data creating unit 801 will be described.

As shown in fig. 1, the learning data creation unit 801 creates supervisory data S7a for the neural network 111 used in the control rule learning unit 802, based on the control result goodness determination estimate value S11 from the control output goodness determination rule execution unit 17, and based on the control operation end operation command S2, the selection control operation amount S8, the control method selection flag S14, and the determination result (control operation amount output availability data S4) of the control output suppression unit.

The supervisory data S7a in this case is the AS-U operation degree 301 and the first intermediate operation degree 302, which are outputs from the output layer of the neural network 111 shown in fig. 7. The learning data creation unit 7 creates supervisory data S7a for the neural network 111 used in the control rule learning unit 802, using the control operation end operation command S2(AS-U operation degree 301, first intermediate operation degree 302) AS the output of the neural network 101, and the #1 to # nAS-U position change amount output, the upper first intermediate shift position change amount output, and the lower first intermediate shift position change amount output AS the selection control operation amount S8.

In describing the operation of the learning data creating unit 801, fig. 14 shows the relationship between the data of each unit and the reference numeral in the control output calculating unit 3 shown in fig. 10. Here, the control operation side operation command S2, which is an output of the neural network 101, representatively shows the AS-U operation degree 301, and data on the positive side of the operation degree is represented by OPref, data on the negative side of the operation degree is represented by OMref, the conversion gain is represented by G, and the control operation amount S3 is represented by Cref.

The positive operation degree data OPref and the negative operation degree data OMref are differentiated by a subtractor 701, and multiplied by a conversion gain G by a multiplier 702 to obtain a control operation amount output Cref. The control operation amount output Cref is supplied to the control output operation method selection unit 18, and the selected operation command value C ″ ref is obtained.

Here, for simplicity, as the output from the output layer of the neural network 101 of the control rule executing unit 10, the degree of operation of random generation by the control operation disturbance generating unit 16 on the positive side and the negative side of the degree of operation is set as the degree of operation random number. The control manipulated variable output SO to the control manipulation end is set as a manipulation command value.

Fig. 15 shows the processing stages and the processing contents in the learning data creation unit 7.

In the first processing stage 71, the operation command value C ″ ref refers to the selection control operation amount S8 as the output value of the control output operation method selection unit 18.

In the next processing stage 72, the operation command value Cref is corrected and set to C' ref based on the control result goodness determination estimation value S11 and the control operation amount output possibility data S4. Specifically, when the control result superiority/inferiority determination estimated value S11 is 0or the control manipulated variable output propriety data S4 is 0, the correction value C' ref of the manipulation instruction value C "ref is set by the following equation 5 when the control result superiority/inferiority determination estimated value S11 is 1 and the control manipulated variable output propriety data S4 is 1, by the following equation 6.

[ numerical formula 5]

IFC”ref＞OTHEN C’ref＝C”ref-Δcref

IFC”ref＜OTHEN C’ref＝C”ref+Δcref

[ numerical formula 6]

IFC”ref＞OTHEN C’ref＝C”ref+Δcref

IFC”ref＜OTHEN C’ref＝C”ref-Δcref

In the processing stage 73, the operation degree correction amount Δ Oref is obtained from [ equation 7] and [ equation 8] based on the corrected operation command value C' ref.

[ number formula 7]

C′ref＝G-((OPref+ΔOref)-(OMref-ΔOref))

[ number formula 8]

In the processing stage 74, the monitoring data OP 'ref, OM' ref for the neural network 111 are obtained by [ equation 9 ].

[ number formula 9]

As described above, in the learning data creation unit 7, as shown in fig. 14, the operation command value correction value C' ref is calculated for the operation command value C ″ ref actually output to the plant 1 to be controlled, based on the control result goodness determination estimate value S11 of the control output goodness determination rule execution unit 17 and the control operation amount output availability data S4 of the control output suppression unit 4.

Specifically, when the control result goodness determination estimate S11 is equal to 1 and the control manipulated variable output propriety data S4 is equal to 1, the manipulated command value is increased by Δ cref in the same direction when it is determined that the manipulation is a good manipulation.

In contrast, when the control result goodness determination estimate S11 is equal to 0or the control manipulated variable output propriety data S4 is equal to 0, the manipulated command value is decreased by Δ cref in the opposite direction when it is determined that the manipulation is bad. Since the conversion gain G is set in advance and is known, the correction amount Δ Oref can be obtained by knowing the values on the positive side and the negative side of the degree of operation. Here, Δ Cref is determined and set to an appropriate value in advance by simulation or the like. Through the above steps, the supervision data OP 'ref and OM' ref used in the control rule learning unit 802 can be obtained by [ expression 9 ].

Note that, although the description has been made with a simple example in fig. 14, in actuality, all the steps are executed with respect to the AS-U operation degree 301 with respect to #1 to # nAS-U and the first intermediate operation degree 302 with respect to the upper first intermediate roller shift and the lower first intermediate roller shift, and the supervision data (AS-U operation degree supervision data, first intermediate operation degree supervision data) of the neural network 111 used in the control rule learning unit 802 is used.

Fig. 16 shows an example of data stored in the learning data database DB 2.

In order to learn the neural network 111, a combination of a plurality of input data S8a and supervisory data S7a is required. Therefore, the supervision data S7a created by the learning data creation unit 7 is combined with the input data S1(S8a) input to the control rule execution unit 10 by the control execution unit 20 and stored in the learning data database DB2 as a set of learning data. The supervision data S7a here is AS-U operation degree supervision data, a first intermediate operation degree. The input data S1(S8a) is the normalized shape deviation 201 and the shape deviation stage.

The plant control system of fig. 1 uses various databases DB1, DB2, DB3, and DB4, but the databases DB1, DB2, DB3, and DB4 are managed and operated in cooperation with the neural network management table TB.

Fig. 17 shows the structure of the neural network management table TB.

The neural network management table TB is divided according to the (B1) board width, the (B2) steel grade, and the specifications a1, a2 regarding the priority of control. As the sheet width (B1), for example, 4 divisions of 3 feet width, meter width, 4 feet width and 5 feet width are used, and as the steel type, 10 divisions of steel type (1) to steel type (10) are used. Further, the specifications a regarding the priority of control are 2 types, i.e., a1 and a 2. In this case, 80 divisions are obtained, and 80 neural networks are used separately according to rolling conditions.

The neural network learning control unit 112 associates learning data, which is a combination of input data and supervisory data as shown in fig. 16, with the corresponding neural network No. in accordance with the neural network management table TB shown in fig. 17, and stores the learning data in the learning data database DB2 as shown in fig. 18.

The control execution unit 20 generates 2-group learning data each time shape control is executed on the control target plant device 1. This is because 2 types of supervisory data are created for the same input data and control output, since the quality determination of the control result is performed using 2 evaluation criteria, that is, the specification a1 and the specification a2, regarding the priority of control. When the supervised data is accumulated to a certain extent (for example, 200 sets) or newly accumulated in the learning data database DB2, the neural network learning control unit 112 instructs the neural network 111 to learn.

In the control rule database DB1, a plurality of neural networks are stored in accordance with the management table TB shown in fig. 17. The neural network learning control unit 112 specifies the neural network No. to be learned, and the neural network selecting unit 113 extracts the neural network from the control rule database DB1 and sets the neural network as the neural network 111.

The neural network learning control unit 112 instructs the input data creation unit 114 and the supervisory data creation unit 115 to extract the input data and the supervisory data corresponding to the respective neural networks from the learning data database DB2, and to use these data to perform learning of the neural network 111. Various methods have been proposed as to the learning method of the neural network, and any method may be used.

When the learning of the neural network 111 is completed, the neural network learning control section 112 writes back the neural network 111 as a result of the learning to the position of the corresponding neural network No. of the control rule database DB1, whereby the learning is completed.

Learning may be performed at regular time intervals (for example, every 1 day) for all the neural networks defined as shown in fig. 17, or learning may be performed only for the neural network No. in which new learning data of a certain degree (for example, 100 sets) are accumulated at that time point.

Next, the operation of the merit judgment rule learning section 22 will be described.

The goodness determination rule learning unit 22 uses time delay data of data used in the control execution unit 20. Here time delay Z^-1Meaning e-TS, indicates that the delay is performed for a predetermined time T.

Since the plant equipment 1 to be controlled has a time response, there is a time delay until the actual result data changes by outputting the SO according to the control manipulated variable. Therefore, learning is performed using actual performance data at a time point when the delay time T has elapsed after the control operation is performed.

In the shape control, it takes several seconds until the shape change is detected by the shape meter after the operation command is output to the AS-U or the first intermediate roller, and therefore it is preferable to set T to about 2 to 3 seconds. Since the delay time also varies depending on the type of the shape detector and the rolling speed, it is preferable to set the optimum time until the change of the control operation end to the shape change to T.

Fig. 19 shows the operation of the control result quality determination unit 6. The shape change goodness determination unit 602 uses a goodness determination evaluation function Jc expressed by equation 10.

[ numerical formula 10]

In the expression 10, ∈ fb (i) is shape deviation actual result data included in the actual result data Si, ∈ last (i) is a previous value of the shape deviation actual result data, and wc (i) is a weighting coefficient in the width direction for quality determination. Here, the weighting coefficients wc (i) for quality determination are set from the quality determination database DB4 in accordance with the specifications a1 and a2 regarding the priority of control. And judging the quality of the control result according to the quality judgment evaluation function Jc.

The upper threshold LCU and the lower threshold LCL are preset on the basis of the threshold condition (LCU is more than or equal to 0 and is more than or equal to LCL). At this time, if the comparison result with the goodness determination evaluation function Jc is Jc > LCU, the control result goodness data S6 is 0 (shape degradation), and if Jc < LCL, the control result goodness data S6 is 1 (shape degradation).

In this way, the weighting coefficient wc (i) in the plate width direction changes depending on the specifications a1 and a2 regarding the priority of control, and thus the merit determination evaluation function Jc differs. Therefore, the determination result of the control result goodness data S6 is also considered to be different. Therefore, the good and bad determination rule learning unit 22 performs the determination of the control result good and bad data S6 for 2 types of specifications a1, a2 regarding the priority of control.

This control result good and bad data S6 is directly used as the supervision data S13a for the neural network 311 used in the good and bad determination rule learning section 31.

Fig. 20 shows an example of data stored in the learning data database DB 6.

In order to learn the neural network 311, a combination of a plurality of input data S12a and supervisory data S13a is required. Therefore, the supervision data S13a (control result goodness data) created by the control result goodness determination unit 6 is combined with the time lag data S12a of the input data S1 (normalized shape deviation 201 and shape deviation phase) input to the control rule execution unit 10 by the control execution unit 20 and stored as a set of learning data in the learning data database DB 6.

In this case, the learning data is stored in the verification data DB7 at a constant rate and can be used for the quality determination rule verification in the quality determination rule accuracy verification unit 34.

The goodness determination rule accuracy verification unit 34 includes a neural network that performs calculation in only one direction, as in the control output goodness determination rule execution unit 17. Then, the quality determination rule accuracy verification unit 34 extracts the test data from the verification data DB7, and calculates an error between the output data obtained by inputting the input data of the test data to the neural network and the output data included in the test data. For example, the goodness determination rule accuracy verification unit 34 calculates the average value of the errors of all the test data and the like as the goodness determination rule accuracy S15 of the goodness determination rule.

Although the plant control system of fig. 1 uses various databases DB5 and DB6, fig. 21 shows a configuration of a neural network management table TB for managing and operating in cooperation with the databases DB5 and DB 6. That is, the management table TB includes a standard management table.

Specifically, as shown in fig. 21, the management table TB is divided according to the (B1) sheet width, the (B2) steel grade, and the specifications a1, a2 regarding the priority of control. As the sheet width (B1), for example, 4 divisions of 3 feet width, meter width, 4 feet width and 5 feet width are used, and as the steel type, 10 divisions of steel type (1) to steel type (10) are used. Further, 2 types of specifications a regarding the priority of control are a1 and a 2. In this case, 80 divisions are obtained, and 80 neural networks are used separately according to rolling conditions.

The neural network learning control unit 312 associates learning data, which is a combination of input data and supervisory data as shown in fig. 20, with the corresponding neural network No. in accordance with the neural network management table TB shown in fig. 21, and stores the learning data in the learning data database DB6 as shown in fig. 22.

Each time the control execution unit 20 executes the shape control of the control

target plant device

1, 2 group learning data is generated. This is because 2 types of supervisory data are created for the same input data and control output, since the quality determination of the control result is performed using 2 evaluation criteria, that is, the specification a1 and the specification a2, regarding the priority of control. When the supervised data is accumulated to a certain extent (for example, 200 sets) or newly accumulated in the learning data database DB6, the neural network learning control unit 312 instructs the neural network 311 to learn.

The quality determination rule database DB5 stores a plurality of neural networks in accordance with the management table TB shown in fig. 21. Then, the neural network learning control unit 312 specifies the neural network No. to be learned, and the neural network selection unit 313 extracts the corresponding neural network from the quality determination rule database DB5 and sets it as the neural network 311. The neural network learning control unit 312 extracts input data and supervisory data corresponding to the corresponding neural network from the learning data database DB6, instructs the input data creation unit 314 and the supervisory data creation unit 315, and performs learning of the neural network 311 using these data. Various methods have been proposed as methods for learning a neural network, and any method may be used.

When the learning of the neural network 311 is completed, the neural network learning control section 312 writes back the neural network 311 as a learning result to the position of the neural network No. of the control rule database DB6, whereby the learning is completed.

Learning is performed at regular time intervals (for example, every 1 day) for all the neural networks defined in the management table TB shown in fig. 21. Alternatively, learning may be performed only for the neural network No. in which new learning data is accumulated to some extent (for example, 100 sets) at that time point.

In addition, by including the rolling result, the steel grade, and the sheet width in the input data of the goodness determination rule, it is possible to learn through 1 neural network including the difference of the goodness determination criterion. In this case, it is not necessary to switch the good/bad determination rule according to the rolling condition when executing the good/bad determination rule.

As described above, in order to improve the control rule of the plant 1 to be controlled, if the control operation having the good control result cannot be learned, the control operation is largely changed. In addition, when the control result is good, a new control operation method is adopted. In addition, when a control operation with a good control result can be learned, the control operation is not changed or is limited to a small change. When the control result for these operations is good, it is effective to take in the control operation as a new control operation method.

Further, by learning a combination of the shape pattern, the control operation, and the merits of the control result based on the real machine data, it is possible to construct a model that can estimate the merits of the control result with higher accuracy than a simulator using a machine model, and it is possible to construct a model that is always suitable for the latest plant state by automatic learning at regular intervals.

Further, the reliability of the control output suppressing function to the plant equipment, which has been performed only by a simple mechanical model in the related art, can be improved by using a model for estimating the quality determination of the control result.

In addition, in the case of the present example, it is possible to suppress the influence of noise included in the plant equipment data by using the model for judging the merits of the estimated control results, and to perform fine adjustment with a small effect as the target of the learning data, with respect to the generation of the control rule learning data which has been conventionally performed in the judgment of the merits of the primary control results. Further, according to this example, by preventing erroneous determination of the control effect, it is possible to suppress variation in the learning data and achieve the effect of stabilizing the control performance.

In addition, the neural network used in the control execution unit 20 is stored in the control rule database DB 1. Here, when the stored neural network performs the initial processing only by the random number, it takes time until the neural network performs learning until the corresponding control can be performed. Therefore, when the control unit is constructed for the control target plant 1, the learning of the control rule is executed in advance by simulation based on the control model of the control target plant 1 found at that time point. Then, by storing the neural network in which the learning in the simulator is completed in the database, the control of the performance to some extent can be executed from the start of the plant equipment to be controlled.

Alternatively, by causing the merit/disadvantage determination rule learning means 22 to learn the merit/disadvantage determination rule based on the performance data of the operation data in the actual plant, the control rule can be learned without controlling the actual plant, and the control of the performance to some extent can be executed before being applied to the plant to be controlled.

Fig. 23 shows a configuration in a case where a control rule evaluation unit 23 that performs evaluation processing of a control rule is provided as the plant control system of the present example.

The control rule evaluation unit 23 includes a control rule goodness determination data collection unit 35, a control rule evaluation data calculation unit 36, a control rule database update unit 37, a control rule evaluation data database DB8, and a control rule evaluation value database DB 9.

The control rule goodness determination data collection section 35 receives the control output goodness determination estimate S9 from the control output goodness determination rule execution section 17, and receives the goodness determination rule accuracy S15 from the goodness determination rule accuracy verification section 34. Then, the control rule goodness determination data collection section 35 saves the control rule goodness determination data S16 in the control rule evaluation data DB8 together with the control rule number used in the control execution unit 20. The control rule good/bad decision data S16 is a control output good/bad decision estimated value S9. However, when the quality determination rule accuracy S15 is equal to or less than a certain value, it is not stored in the database DB 8.

The control rule goodness determination data S16 is new data obtained each time the control execution unit 20 performs an operation using the control output of the control rule, and the obtained control rule goodness determination data S16 is stored in the control rule evaluation data DB 8. In this case, since a large amount of data is stored for each control rule, the control rule evaluation data DB8 determines in advance the upper limit of the data stored in each control rule, and when the upper limit is equal to or greater than a certain value, deletes the old data and stores the new data.

The control rule evaluation data calculation unit 36 collectively extracts the control rule goodness determination data S17 accumulated for each control rule from the control rule evaluation data DB8, and calculates the average value thereof to obtain the control rule evaluation data S18. The obtained average value corresponds to an evaluation value.

The control rule evaluation data S18 calculated by the control rule evaluation data calculation unit 36 is stored in the control rule evaluation value database DB 9. However, when the number of the control rule goodness determination data is less than a certain number, the reliability of the evaluation value is low, and the evaluation result is not stored.

In the database management table TB, the neural network nos. (control rules) used according to the conditions are registered one by one. In contrast, the control rule evaluation value database DB9 manages evaluation values of a plurality of control rules. The control rule database update unit 37 refers to the control rule evaluation value database DB9, compares the control rule evaluation value of the neural network No. (control rule) registered in the database management table TB with the control rule evaluation values of the other control rules applicable to the condition, and updates the control rule having the highest evaluation value among the control rule evaluation values to the neural network No. (control rule) of the database management table TB.

The other parts of the plant control system shown in fig. 23 are configured in the same manner as the plant control system shown in fig. 1. However, in the case of the plant control system shown in fig. 23, since the control rule evaluation means 23 evaluates the control target plant 1 based on the past actual results as the past actual results data of the control target plant 1, the control execution means 20 does not need to actually control the control target plant 1. Specifically, the control output amount S0 does not need to be supplied from the control output suppression unit 4 to the plant 1 to be controlled.

According to the plant control system shown in fig. 23, by setting the control rule to be evaluated in the control rule execution unit 10 and giving the past actual result data as Si, the control rule evaluation value database DB9 can be updated even if the control output to the plant 1 to be controlled is not actually performed.

< modification example >

The present invention is not limited to the examples of the above embodiments, and includes various modifications. For example, the above-described embodiments have been described in detail to facilitate understanding of the present invention, and are not limited to having all of the described configurations.

For example, the plant control system shown in fig. 1 and 23 includes a processing unit that performs processing such as data creation, learning, and control. The control execution means 20, the control method learning means 21, the quality determination rule learning means 22, and the control rule evaluation means 23 shown in fig. 1 and 23 may be configured by a program (software) in which a processor realizes their respective functions, and cause a computer to execute the program. Fig. 24 shows a configuration example of a computer in this case.

That is, as shown in fig. 24, the computer constituting each of the units 20 to 23 includes a CPU (Central Processing Unit) a, a ROM (Read Only Memory) b, and a RAM (Random Access Memory) c, each of which is connected to a bus. The computer is provided with a nonvolatile memory d and a network interface e.

The CPUa is an arithmetic processing unit that reads out and executes program codes of software that executes processing in each of the units 20 to 23 from the ROMb. Variables, parameters, and the like generated during the arithmetic processing are temporarily written in the RAMc. The nonvolatile memory d stores programs executed by the units 20 to 23, data of databases, and the like, using a large-capacity information storage unit such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive).

The units 20 to 23 may be constituted by different computers, respectively, but the programs may be installed in a small number of computers such as 1 computer and executed at the same time.

The Network Interface e transmits and receives data to and from other units and the control target plant 1 using, for example, an NIC (Network Interface Card) or the like.

Information such as a program for realizing each processing function in this case can be stored in a recording medium such as a memory, an IC card, an SD card, or an optical disc, in addition to the nonvolatile memory d such as an HDD or an SSD.

In addition, part or all of the functions performed by each of the units 20 to 23 may be realized by hardware such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit).

In the block diagrams shown in fig. 1, 23, and the like, the control lines and the information lines are only portions that are considered necessary for the description, and not necessarily all the control lines and information lines on the product. In practice, it is also possible to consider almost all structures connected to one another.

In the above-described example of the embodiment, the plant 1 to be controlled is applied to the sendzimir mill, but the present invention can be applied to control of various other plant. The control rule applied to the sendzimir rolling mill is also an example, and the present invention is not limited to the example of the above embodiment.

Claims

1. A plant control system for executing control by recognizing, for a plant to be controlled, a pattern of a combination of actual performance data of the plant to be controlled,

the plant control system includes:

a control method learning unit that learns a combination of actual performance data and a control operation of the control target plant equipment;

a control execution unit that executes control of the plant equipment to be controlled based on a combination of the actual performance data and the control operation learned by the control method learning unit; and

a goodness determination rule learning unit that learns a combination of actual performance data and a control operation of the control target plant and a goodness of a control result,

the control execution unit includes:

a control rule execution unit that provides a control output in accordance with a combination of the actual performance data of the plant equipment to be controlled and the determination of the control operation;

a control output quality determination rule execution unit that determines the quality of the control output in accordance with a combination of actual performance data of the plant equipment to be controlled, control operation, and determination of the quality of a control result;

a control output suppressing unit that prevents the output of the control output to the plant equipment to be controlled when it is determined that the performance data of the plant equipment to be controlled is deteriorated when the control output is output to the plant equipment to be controlled, using the goodness determination by the control output goodness determination rule executing unit and the simulation data using the control simulator,

the quality determination rule learning means includes:

a control result quality determination unit that determines the quality of the control result after a time delay until the control effect is expressed in the actual performance data when the control execution means outputs the control output to the plant to be controlled; and

a goodness determination rule learning unit that learns the goodness of the control result, the achievement data, and the control output in the control result goodness determination unit as learning data,

the control method learning unit is provided with:

a learning data creation unit that obtains supervision data using the control output and the control output in the control output quality determination rule execution unit; and

and a control rule learning unit that learns the actual result data and the supervision data as learning data.

2. The plant control system of claim 1,

the control method learning means performs learning to obtain different combinations of actual performance data and control operations for a plurality of control targets depending on the state of the plant to be controlled,

the obtained combination of the actual performance data and the control operation is used as the determined combination of the actual performance data and the control operation of the control target plant equipment in the control rule execution unit.

3. The plant control system of claim 1,

the control output goodness determination rule execution section maintains a combination of the actual performance data of the control target plant equipment, the control operation, and the determination of the goodness of the control result as a first neural network,

the goodness determination rule learning section holds a combination of goodness of the actual performance data, the control operation, and the control result as a second neural network,

the second neural network obtained as a result of learning in the goodness determination rule learning section is used as the first neural network in the goodness determination rule execution section.

4. The plant control system of claim 1,

the good/bad judgment rule learning means includes a good/bad judgment rule accuracy verification section,

the control output suppression unit may be configured to change a reference of the output suppression using the goodness of the control result in the control output suppression unit, using the goodness determination rule accuracy generated by the goodness determination rule accuracy verification unit.

5. The plant control system according to any one of claims 1 to 4,

the plant control system is further provided with a control rule evaluation unit,

the control rule evaluation unit includes:

a control rule good/bad determination data collection unit that stores, in a database, good/bad determination data of the good/bad determination rule execution unit of the control execution unit and a result of accuracy verification of the good/bad determination rule by the good/bad determination rule learning unit; and

a control rule evaluation data calculation unit that calculates control rule evaluation data based on the good/bad determination data accumulated in the database and the result of accuracy verification of the good/bad determination rule,

the plant control system executes the evaluation of the control rule used by the control execution unit without outputting the evaluation to the plant to be controlled.

6. A plant control method for identifying a pattern of a combination of performance data of a plant to be controlled with respect to the plant to be controlled and controlling the plant to be controlled by a computer, characterized in that,

as the computer-executed process, there are included:

a control method learning process of learning a combination of actual performance data and a control operation of the control target plant equipment;

a control execution process of executing control of the control target plant equipment based on a combination of the actual performance data and the control operation learned by the control method learning process; and

a goodness determination rule learning process of learning a combination of actual performance data and a control operation of the control target plant and a goodness of a control result,

the control execution process includes:

a control rule execution process of providing a control output in accordance with a determined combination of the actual performance data and the control operation of the control target plant equipment;

a control output quality determination rule execution process of performing quality determination of the control output based on a combination of actual performance data of the control target plant equipment, control operation, and determination of quality of a control result;

new search operation amount calculation processing of calculating an operation amount for new operation search based on the goodness determination of the control output goodness determination rule execution processing; and

a control output suppression process of preventing a control output from being output to the plant device to be controlled when it is determined that performance data of the plant device to be controlled is deteriorated when the control output is output to the plant device to be controlled using a goodness determination for performing a process based on the control output goodness determination rule and simulation data using a control simulator,

the good and bad judgment rule learning process includes:

a control result goodness determination process of determining goodness of a control result after a time lag until a control effect is expressed to actual performance data in a case where a control output is output to the control target plant by the control execution process; and

a goodness determination rule learning process of learning, as learning data, the goodness of the control result, the performance data, and the control output in the control result goodness determination process,

the control method learning process includes:

learning data creation processing of performing a quality determination of the control output of the processing by the control output quality determination rule and obtaining supervision data using the control output; and

and a control rule learning process for learning the actual performance data and the supervision data as learning data.

7. A computer-readable recording medium storing a program for causing a computer to execute plant equipment control by identifying a pattern of a combination of performance data of a plant equipment to be controlled with respect to the plant equipment to be controlled,

the program causes the computer to execute:

a control method learning step of learning a combination of actual performance data and a control operation of the control target plant equipment;

a control execution step of executing control of the plant equipment to be controlled based on a combination of the actual performance data and the control operation learned by the control method learning step; and

a quality determination rule learning step of learning a combination of actual performance data and a control operation of the control target plant and a quality of a control result,

the control execution step includes:

a control rule execution step of providing a control output based on a determined combination of performance data and control operation of the control target plant equipment;

a control output quality determination rule execution step of performing quality determination of the control output based on a combination of actual performance data of the control target plant equipment, control operation, and determination of quality of a control result;

a new search operation amount calculation step of calculating an operation amount for searching for a new operation based on the determination of the quality in the control output quality determination rule execution step; and

a control output suppressing step of preventing the output of the control output to the plant device to be controlled when it is determined that the performance data of the plant device to be controlled is deteriorated when the control output is output to the plant device to be controlled using the goodness determination of the control output goodness determination rule executing step and the simulation data using the control simulator,

the step of learning the goodness determination rule includes:

a control result quality determination step of determining quality of a control result after a time delay until a control effect is expressed to actual performance data when a control output is output to the plant to be controlled by the control execution step; and

a quality determination rule learning step of learning the quality of the control result in the control result quality determination step, the performance data, and the control output as learning data,

the control method learning step includes:

a learning data creation step of obtaining supervisory data by using the control output quality determination and the control output of the control output quality determination rule execution step; and

and a control rule learning step of learning the actual performance data and the supervision data as learning data.