JP6952018B2 - Control device and control method - Google Patents

Description

The present invention relates to a technique for performing real-time feedback control using artificial intelligence such as a neural network.

Conventionally, in various plants, plant control based on various control theories has been carried out in order to obtain a desired control result by the control.

As an example of a plant, for example, in rolling mill control, fuzzy control and neuro-fuzzy control have been applied as control theories for shape control for controlling the wavy state of a plate as an example of control. Fuzzy control is applied to shape control using coolant, and neuro-fuzzy control is applied to shape control of a sentimia rolling mill. Of these, the shape control to which the neuro-fuzzy control is applied is similar to the difference between the actual shape pattern detected by the shape detector and the target shape pattern and the preset reference shape pattern, as shown in Patent Document 1. The ratio is obtained, and the control output amount for the operation end is obtained from the similar ratio according to the control rule expressed by the control operation end operation amount for the preset reference shape pattern. Hereinafter, as a conventional technique, shape control of a sendimia rolling mill using neuro-fuzzy control will be used.

FIG. 1 shows the shape control of the Sendimia rolling mill described in FIG. 1 of Patent Document 1. Neuro-fuzzy control is used to control the shape of the Sendimia rolling mill. In this example, the pattern recognition mechanism 51 recognizes the shape pattern from the actual shape detected by the shape detector 52, and calculates which of the preset reference shape patterns the actual shape is closest to. The control calculation mechanism 53 executes control using a control rule composed of a control operation end operation amount for a preset shape pattern as shown in FIG. More specifically, FIG. 2 shows that in the pattern recognition mechanism 51, the difference (Δε) between the actual shape detected by the shape detector 52 and the target shape (εref) is a shape pattern (ε) of 1 to 8. The closest control method is calculated, and the control calculation mechanism 53 selects and executes any of the control methods 1 to 8.

However, in the method of Patent Document 1, in order to verify the control rule, the operator may manually perform the operation during rolling to verify the control rule, but the shape may change unexpectedly. That is, there may be a case where the control rule determined as described above does not conform to the reality. This is due to insufficient examination of mechanical characteristics and changes in the operating conditions and machine conditions of the rolling mill, but it should be considered to verify whether the preset control rules are the best rules one by one. There are many conditions and it is difficult. Therefore, once the control rule is set, it is often left as it is unless there is a problem.

When the control rules do not conform to the reality due to changes in operating conditions or the like, the control rules are fixed, and it becomes difficult to obtain control accuracy above a certain level. Further, once the shape control is operated, the operator does not perform the manual operation (it becomes a disturbance for the control), so that it is difficult to find a new control rule by the manual intervention of the operator. Furthermore, when rolling a rolled material of a new standard, it is difficult to set control rules according to the material.

As described above, in the conventional shape control, since the control is performed by using the preset control rule, there is a problem that it is difficult to modify the control rule.

In order to solve this problem, as shown in Patent Document 2, the control rule is randomly changed while performing shape control, and the rule for improving the shape is learned.
1) Discover new control rules while controlling the shape during rolling.
2) New control rules are not predictable in advance, and control rules that could not be predicted at all may be optimal. Therefore, operate the control operation end at random and find it while looking at the control results for it. go.
I have realized that.

Japanese Patent No. 2804161 Japanese Patent No. 4003733

In the above-mentioned prior art, a representative shape is set in advance as a reference shape pattern, and control is performed based on a control rule indicating a relationship with a control operation end operation amount with respect to the reference waveform pattern. Also in the learning of the control rule, it is related to the control operation end operation amount with respect to the reference waveform pattern, and the typical reference shape pattern defined in advance is used as it is. Therefore, there is a problem that the shape control reacts only to a specific shape pattern.

The standard shape pattern is determined by human beings in advance based on their knowledge of the target rolling mill and the experience of accumulating shape results and manual intervention operations, but all shapes generated in the target rolling mill and the material to be rolled It is difficult to cover. Therefore, when a shape different from the reference shape pattern is generated, the control by the shape control is not performed and the shape deviation remains unsuppressed, or it is erroneously recognized as a similar reference shape pattern and erroneously controlled. In some cases, the shape may be deteriorated by performing an operation.

Therefore, in the conventional shape control, there is a problem that there is a limit to the improvement of the control accuracy because the control rule is learned and the control is performed by using the preset reference shape pattern and the control rule for the reference shape pattern. rice field.

In order to solve this problem, it is a plant control device that recognizes the pattern of the combination of the actual data of the controlled plant and executes the control for the controlled plant. It is equipped with a control method learning device that learns combinations and a control execution device that controls the controlled plant according to the combination of the learned actual data and control operation. The control execution device is the actual data and control operation of the controlled plant. The control rule execution unit that gives control output according to the specified combination and the control output output by the control rule execution unit are judged, and the control method learning device is notified that the actual data and the control operation are incorrect. If the control output judgment unit and the control output judgment unit output the control output to the control target plant and determine that the actual data of the control target plant deteriorates, prevent the control output from being output to the control target plant. The control method learning device is equipped with a control output suppression unit, and when the control execution device actually outputs the control output to the controlled plant, the actual data is displayed after a time delay until the control effect appears in the actual data. The control result quality judgment unit that determines whether the control result is better or worse than before the control, the control result quality judgment unit in the control result quality judgment unit, and the teacher data using the control output are used. It is equipped with a learning data creation unit to obtain and a control rule learning unit that learns actual data and teacher data as learning data. The feature is that separate performance data and control operation combinations are obtained, and the obtained performance data and control operation combination is used as a defined combination of control target plant performance data and control operation in the control rule execution unit. It is conceivable to use a plant control device.

At this time, it is very important that the evaluation function used for determining the quality of the control result is appropriate. However, when determining the evaluation function, the designer of the control device subjectively determines it while conducting an interview survey with the operation engineers and operators of the controlled plant and confirming the actual operation of the plant. In many cases, it is unclear whether the settings are truly appropriate.

As an example, consider the shape control of a rolling mill. In the shape control of the rolling mill, it is ideal that the target shape and the actual shape match in the entire plate width direction. However, in reality this is often not the case. Therefore, in actual work, it is common to emphasize a specific region of the plate and control the actual shape so as to match the target shape in that region. As an evaluation function for evaluating the shape of the plate, an evaluation function is used in which the shape deviation (= actual shape-target shape) in each part in the plate width direction is weighted for each part in the plate width direction.

In the rolling mill, the control operation end for the shape of the end portion (plate end portion) in the plate width direction is separate from the control operation end for the portion (center portion) other than the end portion (plate end portion). However, they often influence each other. Further, since the plate end portion is not constrained from both sides unlike the central portion, the shape deviation often becomes large. When control is applied to the plate end portion in the plate width direction, the effect extends to the central portion and the shape of the central portion deteriorates, and vice versa. As described above, it is difficult to control the shapes of the plate edge portion and the central portion at the same time so as to match the target value. In many cases, the operator gives priority to either the edge portion or the center portion of the plate to perform manual control.

When the evaluation function applied in the quality judgment of the control result evaluates differently from the operator's idea, the operator cancels the operation from the shape control by the control device and performs the manual operation according to his / her own idea. It will be. In that case, the shape control by the control device and the manual operation performed by the operator conflict with each other. As a result, it is conceivable that the operator turns off the shape control from the control device, which is an obstacle to the manual operation by the operator. If this is repeated, there is a concern that the operator will not turn on the shape control by the control device from the beginning.

If the evaluation function applied to the quality judgment of the control result is evaluated in accordance with the operator's idea, not only the conflict between the control by the control device and the manual operation of the operator is reduced, but also the operator can perform the evaluation. It is also expected that the number of manual operations will be reduced, the load on the operator will be reduced, and the accuracy of shape control will be improved.

An object of the present invention is to provide a technique that enables control based on an appropriate pass / fail judgment of a control result.

The control device according to the present disclosure is a control device that controls a control target, and is a control execution device that gives a control output to the control target according to a given control rule, and the control target using a designated evaluation function. A control method in which the given control output is evaluated, training data is created using the evaluation result, the control rule is constructed by learning the training data, and the control rule is given to the control execution device. A learning device and a plurality of evaluation functions are held in advance, one of the plurality of evaluation functions is selected based on the control state of the controlled object, and the selected evaluation function is used as the control method learning device. It has an evaluation function setting unit specified in.

According to the present disclosure, it is expected that control based on an appropriate pass / fail judgment of the control result can be executed.

It is a figure which shows the shape control of the Sendimia rolling mill described in FIG. 1 of Patent Document 1. FIG. It is a figure which shows the control rule which is composed of the control operation end operation amount for a shape pattern. It is a figure which shows the outline of the plant control apparatus which concerns on Example. It is a figure which shows the specific example of the control rule execution part 10 which concerns on embodiment. It is a figure which shows the specific example of the control rule learning part 11 which concerns on embodiment. It is a block diagram which shows the internal structure of the evaluation function setting part 17. It is a figure which shows the structure of the neural network when it is used for the shape control of a Sendimia rolling mill. It is a figure for demonstrating a shape deviation and a control method. It is a figure which shows the outline of the control input data creation unit 2. It is a figure which shows the outline of the control output calculation unit 3. It is a figure which shows an example of the transition of the rolling speed of a rolling mill. It is a figure which shows an example of the evaluation function DB DB5. It is a figure for demonstrating the operation outline of the evaluation function selection method learning part 173. It is a figure for demonstrating the operation outline of the evaluation function learning part 174. It is a figure which shows the outline structure of the evaluation function learning part 174. It is a figure for demonstrating the outline of the control output determination part 5. It is a figure for demonstrating the operation outline of the control quality determination part 6. It is a figure for demonstrating the operation outline of the learning data creation part 7. It is a figure which shows the processing stage and processing content in a learning data creation unit 7. It is a figure which shows the example of data stored in the learning data database DB2. It is a figure which shows an example of the neural network management table TB. It is a figure which shows an example of the learning data database DB2.

First, the findings in the present invention and the background to the present invention will be described by taking shape control of a rolling mill as an example.

First, the following three things are required to solve the above problems.

(1) Rather than setting the reference shape pattern and the control operation for it separately in advance and learning the control operation method, the combination of the shape pattern and the control operation is learned and the control operation is performed using it. do

(2) New control rules are not predictable in advance, and control rules that could not be predicted at all may be optimal. Therefore, the control operation end is randomly operated and found while looking at the control results for it. To go

(3) Regarding the quality of the control result, the evaluation function is selected according to the state of the rolling mill, and a suitable control rule can be selected.

In order to realize these three, it is preferable to change the control operation so that the control result is improved while changing the combination of the shape pattern and the control operation used for the shape control. For that purpose, the combination of the shape pattern and the control operation suitable for the shape pattern is learned by artificial intelligence such as a neural network, and the output of the control operation for the shape pattern generated by the rolling mill is changed by the artificial intelligence. It is good to go.

If the control operation is changed while performing shape control on the rolling mill in operation, an erroneous control output may be output, the shape of the plate may deteriorate, and an operation abnormality such as plate breakage may occur. When a plate break occurs, it takes time to replace the roll used in the rolling mill, and the material to be rolled during rolling is wasted, resulting in great damage. Therefore, it is necessary not to output an erroneous control output to the rolling mill as much as possible. Therefore, it is preferable to change the evaluation function for determining the quality of the shape according to the rolling state.

The rolling state is a state related to rolling in which the rolling mill to be controlled is placed. If the control target is not limited to the rolling mill, the rolling state can be generalized and called the control state. The rolling state can be determined by various parameters such as a control operation applied to the rolling mill, a rolling mill state, and a rolling state by the rolling mill. In the present embodiment, as an example, the rolling state is determined by the rolling speed.

From the above, in this embodiment, in order to realize this, the quality of the control operation output by the neural network is verified by using, for example, a simple model of a rolling mill, and the output that is considered to have a clearly deteriorated shape is obtained. , Prevents shape deterioration by not outputting to the control operation end of the rolling mill. At this time, with respect to the neural network, learning is performed assuming that the control operation for the shape pattern is incorrect.

Since there is a possibility that the verification method itself of the quality of the control operation is wrong, it is assumed that the control operation output of the neural network, which is judged to be wrong with a certain probability, is also output to the control operation end of the rolling mill. It is possible to learn about the combination of the outer shape pattern and the control operation.

Hereinafter, examples of the present invention will be described in detail with reference to the drawings.

FIG. 3 shows an outline of the plant control device according to the embodiment. The plant control device of FIG. 3 inputs the control target plant 1 and the actual data Si from the control target plant 1 and controls the control operation amount output SO determined according to the control rule (neural net) as illustrated in FIG. Control method learning in which the control execution device 20 that is given to and controlled by the target plant 1 and the actual data Si from the control target plant 1 are input and learned, and the learned control rules are reflected in the control rules in the control execution device 20. It is composed of an apparatus 21, a plurality of database DBs (DB1 to DB3), and a management table TB of the database DB.

The control execution device 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, and a control operation disturbance generation unit 16 as main elements. ing.

Of these, in the control execution device 20, first, the input data S1 of the control rule execution unit 10 is created by using the control input data creation unit 2 from the actual data Si of the rolling mill which is the control target plant 1. The control rule execution unit 10 creates a control operation end operation command S2 from the control target actual data Si by using a neural network (control rule) expressing the relationship between the control target actual data Si and the control operation end operation command S2. do. The control output calculation unit 3 calculates the control operation amount S3 to the control operation end based on the control operation end operation command S2. As a result, the control operation amount S3 is created using the neural network according to the actual data Si of the controlled plant 1.

Further, in the control output determination unit 5 in the control execution device 20, whether or not the control operation amount can be output to the control operation end by using the actual data Si from the control target plant 1 and the control operation amount S3 from the control output calculation unit 3. The data S4 is determined. The control output suppression unit 4 determines whether or not the control operation amount S3 can be output to the control operation end according to the control operation amount output availability data S4, and gives the permitted control operation amount S3 to the controlled target plant 1. Output as control operation amount output SO. As a result, the control operation amount S3 determined to be abnormal is not output to the controlled plant 1. The control operation disturbance generation unit 16 generates a disturbance and gives it to the controlled plant 1 for the purpose of verifying the plant control device.

The control execution device 20 configured as described above refers to the control rule database DB1 and the output determination database DB3 for the purpose of executing the processing, as will be further described later. The control rule database DB 1 is accessiblely connected to both the control rule execution unit 10 in the control execution device 20 and the control rule learning unit 11 in the control method learning device 21 described later. The control rule (neural network) as a learning result in the control rule learning unit 11 is stored in the control rule database DB1, and the control rule execution unit 10 refers to the control rule stored in the control rule database DB1. The output determination database DB 3 is accessibly connected to the control output determination unit 5 in the control execution device 20.

FIG. 4 shows a specific example of the control rule execution unit 10 according to this embodiment. The control rule execution unit 10 inputs the input data S1 created by the control input data creation unit 2 and gives the control operation end operation command S2 to the control output calculation unit 3. The control rule execution unit 10 includes a neural network 101, and the neural network 101 basically defines a control operation end operation command S2 by the method of Patent Document 1 as illustrated in FIG. In the present invention, the control rule execution unit 10 further includes a neural network selection unit 102, and by referring to the control rules stored in the control rule database DB1, the optimum control rules as the control rules in the neural network 101 Select and execute. As described above, in the control rule execution unit 10 of FIG. 4, a necessary neural network is selected and used from a plurality of neural networks divided by an operator group and a control purpose. It is preferable that the control rule database DB1 also includes the actual data (data of the operation team, etc.) Si that allows the selection of the neural network and the pass / fail judgment criteria as the data from the controlled plant 1. In addition, since there is a relationship that when a neural network is executed, it becomes a control rule, in this specification, the neural network and the control rule are not distinguished and are used in the same meaning.

Returning to FIG. 3, the control method learning device 21 learns the neural network 101 used by the control execution device 20. When the control execution device 20 outputs the control operation amount output SO to the controlled target plant 1, it takes time for the control effect to actually appear as a change in the actual data Si. Therefore, learning is performed using data that is delayed by that time. In FIG. 3, Z ^-1 represents an appropriate time delay function for each data.

The control method learning device 21 is composed of a control result quality determination unit 6, a learning data creation unit 7, a control rule learning unit 11, and an evaluation function setting unit 17 as main elements.

Of these, the control result pass / fail determination unit 6 improves the actual data Si by using the actual data Si from the controlled plant 1, the actual data previous value Si0, and the evaluation function set by the evaluation function setting unit 17. It is determined whether the change is in the direction or the change in the worse direction, and the control result good / bad data S6 is output.

In the learning data creation unit 7 in the control method learning device 21, input data such as the control operation end operation command S2, the control operation amount S3, and the control operation amount output availability data S4 created by the control execution device 20 are input for the same time. Using the data delayed by the time and the control result pass / fail data S6 from the control result pass / fail determination unit 6, new teacher data S7a used for learning the neural network is created and given to the control rule learning unit 11. The teacher data S7a corresponds to the control operation end operation command S2 output by the control rule execution unit 10, and the learning data creation unit 7 uses the control result quality data S6 given by the control result quality determination unit 6. It can be said that the data obtained by estimating the control operation end operation command S2 output by the control rule execution unit 10 is obtained as new teacher data S7a.

FIG. 5 shows a specific example of the control rule learning unit 11 according to this embodiment. The control rule learning unit 11 includes an input data creation unit 114, a teacher data creation unit 115, a neural network processing unit 110, and a neural network selection unit 113 as main components. Further, the control rule learning unit 11 obtains new teacher data S7a from the learning data creation unit 7 from the data S8a in which the input data S1 from the input data creation unit 2 is delayed in time as an external input, and the control rule. Refer to the data accumulated in the database DB1 and the learning data database DB3.

In the control rule learning unit 11, the input data S1 is taken into the neural network processing unit 110 via the input data creation unit 114 after appropriate time delay compensation.

Further, in the control rule learning unit 11, the new teacher data S7a from the learning data creation unit 7 is the total teacher data S7c including the past teacher data S7b stored in the learning data database DB3 in the teacher data creation unit 115. Is given to the neural net processing unit 110. These teacher data S7a and S7b are appropriately stored in the learning data database DB3 and used.

Similarly, the input data S8a from the control input data creation unit 2 is the total input data S8c including the past input data S8b stored in the learning data database DB3 in the input data creation unit 114, and is the neural net processing unit. Given to 110. These input data S8a and S8b are appropriately stored in the learning data database DB3 and used.

The neural network processing unit 110 is composed of a neural network 111 and a neural network learning control unit 112, and the neural network 111 includes input data S8c from the input data creation device 114, teacher data S7c from the teacher data creation unit 115, and the like. The control rule (neural network) selected by the neural network selection unit 113 is taken in, and the finally determined neural network is stored in the control rule database DB1.

The neural network learning control unit 112 controls the input data creation device 114, the teacher data creation unit 115, and the neural network selection unit 113 at appropriate timings, obtains the input of the neural network 111, and outputs the processing result. Control rule Control is performed so that it is stored in the database DB1.

Here, the neural network 101 in the control execution device 20 of FIG. 4 and the neural network 111 in the control method learning device 21 of FIG. 5 are both neural networks of the same concept. The basic conceptual differences in using them will be explained below.

First, the neural network 101 in the control execution device 20 is a neural network having predetermined contents, and obtains a control operation end operation command S2 as a corresponding output when input data S1 is given, so to speak, in one direction. It is a neural network used for the processing of. On the other hand, the neural network 111 in the control method learning device 21 satisfies this input / output relationship when the input data S1 and the input data S8c and the teacher data S7c for the control operation end operation command S2 are set as learning data. This is for finding a neural network by learning.

The concept of basic processing in the control method learning device 21 configured as described above is as follows. First, when the content of the control operation amount output availability data S4 is "OK", the control operation amount output SO is output to the controlled plant 1, and the content of the control result quality data S6 is "good" (actual data Si is improved). In the case of (change in direction), it is determined that the control operation end operation command S2 output by the control rule execution unit 10 is correct, and learning data is created so that the output of the neural network becomes the control operation end operation command S2.

On the other hand, the content of the control operation amount output availability data S4 is "No", or the content of the control operation amount output SO is output to the controlled plant 1 and the content of the control result quality data S6 is "No" (actual data Si becomes worse). In the case of (change in direction), it is determined that the control operation end operation command S2 output by the control rule execution unit 10 is incorrect, and learning data is created so that the output of the neural network is not output. At this time, the neural network output is configured so that two types of outputs, + direction and-direction, are output to the same control operation end as the control output, and the control operation end operation command S2 on the output side is not output. Create training data as in.

Further, in the control rule learning unit 11 illustrated in FIG. 5, as a result of data processing by the neural network learning control unit 112, the processing is performed as follows. Here, first, the control rule execution unit 10 uses learning data that is a combination of S8c in which the input data S1 to the control execution device 20 is delayed by time and the teacher data S7c created by the teacher data creation unit 115. The learning of the neural network 101 used is carried out. Actually, the same neural network 111 as the neural network 101 of the control rule execution unit 10 is provided in the control rule learning unit 11, and the operation test is performed under various conditions to learn the response at that time, and the result of learning is better. The control rule that has been confirmed to occur is obtained. Since it is necessary to perform learning using a plurality of learning data, a plurality of past learning data are extracted from the learning data database DB2 that stores the learning data created in the past, and learning is performed. At the same time, the learning data of this time is stored in the learning data database DB2. Further, the learned neural network is stored in the control rule database DB1 for use by the control rule execution unit 10.

In the learning of the neural net, each time new learning data is created, the past learning data may be used together for learning, or after the learning data is accumulated to some extent (for example, 100 pieces), the past learning is performed. You may learn by using the data together.

In addition, the control result pass / fail determination unit 6 performs a pass / fail determination using the evaluation function set by the evaluation function setting unit 17. The judgment result of the quality of the control result differs depending on the evaluation function used. Therefore, each neural network corresponding to a plurality of evaluation functions is created. Teacher data is created and learned by each evaluation function for the same input data. As a result, a plurality of teacher data are created for one input data and used for learning the neural network corresponding to each teacher data. It is possible to learn a neural network corresponding to multiple evaluation functions at the same time. Here, the plurality of evaluation functions are, for example, in the case of shape control, a plurality of control target items (for example, a plurality of control target items (for example, plate edge portion, center portion, asymmetric portion, etc.)) to be preferentially controlled in the plate width direction. It is an evaluation function used for each policy, such as which of the plate thickness, tension, rolling load, etc.) should be preferentially controlled.

When this embodiment is applied, it is conceivable that once the neural network 101 used in the control rule execution unit 10 is learned, a new control operation will not be executed. Therefore, the control operation disturbance generation unit 16 randomly generates a new operation method in a timely manner, and executes the control operation in addition to the control operation amount S3 to learn the new control method.

Hereinafter, as an example, the details of this plant control method will be described for shape control in a Sendimia rolling mill as shown in Patent Document 1. The shape control will be described assuming that the following specifications A and B are adopted.

Specification A is a specification for the evaluation function, and has information on the priority in the plate width direction. For example, in shape control, it is often difficult to control the plate thickness or the like to a target value over the entire plate width direction due to mechanical characteristics. Therefore, evaluation functions A1 to AN (N is the maximum number of evaluation functions to be set) corresponding to the following plurality of policies are provided in the plate width direction.

The evaluation function is defined so that the better the evaluation, the smaller the value. For example, the squared average of control deviations, the maximum value-the minimum value, and the like.

＜Ａ２：中央部を優先し、中央部の重み付けを重くした評価関数を用いる。＞

＜Ａ３：板端部の、伸び方向は許容する。＞

＜Ａ４．板端部の、張り方向は許容する。＞

＜Ａ５．板端部がデッドバンド内の場合は許容する。＞

＜Ａ６．最大値−最小値＞

Here, as an example, it is assumed that the following six types of policies and evaluation functions A1 to A6 are used.
<A1: An evaluation function is used in which priority is given to the plate edge and the weight of the plate edge is increased. ＞

<A2: An evaluation function that gives priority to the central part and makes the weight of the central part heavier is used. ＞

<A3: The extension direction of the plate edge is allowed. ＞

<A4. The tension direction of the plate edge is allowed. ＞

<A5. If the plate edge is in the dead band, it is allowed. ＞

<A6. Maximum value-minimum value>

FIG. 6 is a block diagram showing the internal configuration of the evaluation function setting unit 17. The evaluation function setting unit 17 includes an evaluation function manual setting unit 171, an evaluation function selection unit 172, an evaluation function selection method learning unit 173, and an evaluation function learning unit 174. The evaluation function setting unit 17 executes the following processing related to the evaluation function in conjunction with the evaluation function DB DB5.

<Process 17-1: Evaluation function setting>
The evaluation function manual setting unit 171 sets the evaluation function. This is a process of formulating and setting the way of thinking about the shape of the operation engineer or the operator in advance.
<Process 17-2: Selection of evaluation function>
The evaluation function selection unit 172 selects an evaluation function to be used by the control execution device 20 according to the rolling state.
<Process 17-3: Learning how to select the evaluation function>
Evaluation function selection method The learning unit 173 learns so that the evaluation function according to the rolling state is selected from the rolling state and the manual operation results of the operator.
<Process 17-4: Learning of the evaluation function itself>
Since the evaluation function preset manually is not always correct, the evaluation function automatic learning unit 174 learns the evaluation function itself. The evaluation function to be learned here is called a learning evaluation function. When learning progresses to some extent, it becomes possible to evaluate using the learning evaluation function. In that case, the learning evaluation function may be used for evaluation as an evaluation function.

Specification B is a specification for dealing with a condition that is known in advance. As an example, since the relationship between the shape pattern and the control method changes under various conditions, it is conceivable that, for example, it is necessary to classify the specification B1 as the plate width and the specification B2 as the steel type. By changing each of the above, the degree of influence on the shape of the shape operation end changes.

In this example, the controlled plant 1 is a Sendimia rolling mill, and the actual data is the actual shape. The Sendimia rolling mill is a rolling mill having a cluster roll for cold rolling a hard material such as stainless steel. In the Zendimia rolling mill, a work roll having a small diameter is used for the purpose of applying a strong rolling reduction to a hard material. Therefore, it is difficult to obtain a flat steel plate. As a countermeasure, the structure of the cluster roll and various shape control units are adopted. In general, the upper and lower first intermediate rolls of a sendimia rolling mill have a one-sided taper so that they can be shifted, and also have six split rolls and two rolls called AS-U on the upper and lower sides. In the example described below, the shape detector detection data is used as the shape actual data Si, and the shape deviation, which is the difference from the target shape, is used as the input data S1. The control operation amount S3 is the AS-U of # 1 to #n and the roll shift amount of the upper and lower first intermediate rolls.

FIG. 7 shows the configuration of the neural network when used for shape control of the Sendimia rolling mill. A neural network may be abbreviated as a neural network. Here, the neural network is the neural 101 for the control rule execution unit 10, and the neural network shown in the neural network 111 for the control rule learning unit 11, but the structure is the same. ..

In the case of shape control of the Sendimia rolling mill shown in this embodiment, the actual data Si from the controlled plant 1 is the data of the shape detector (here, the shape deviation which is the difference between the actual shape and the target shape is output. It is the actual data of the Sendimia rolling mill including), and the control input data creation unit 2 obtains the standardized shape deviation 201 and the shape deviation step 202 as the input data S1. As a result, the input layers of the neural networks 101 and 111 are composed of the normalized shape deviation 201 and the shape deviation step 202. In FIG. 7, the shape deviation step 202 is input to the neural network input layer, but the neural network may be switched according to the step.

Further, the output layer is composed of an AS-U operating degree 301 and a first intermediate operating degree 302 in accordance with the AS-U and the first intermediate roll, which are the shape control operation ends of the Sendimia rolling mill. For AS-U, the degree of operation is the AS-U opening direction (the direction in which the roll gap (distance between the upper and lower working rolls of the rolling mill) opens) and the AS-U closing direction (the direction in which the roll gap closes). I have about AS-U. Regarding the first intermediate roll, the first intermediate roll opening direction (the direction in which the first intermediate roll operates outward from the center of the rolling mill) and the first intermediate roll closing direction (the first intermediate roll is on the center side of the rolling mill). The direction of movement toward the upper and lower first intermediate rolls). For example, if the shape detector has 20 zones and the shape deviation stage 202 is set to 3 stages (large, medium, small), the input layer has 23 inputs. Further, assuming that the AS-U saddles are 7 and the upper and lower first intermediate rolls can be shifted in the plate width direction, the output layer has 14 AS-U operation degrees 301 and 4 1 intermediate operation degrees, for a total of 18 pieces. It becomes. The number of layers in the middle layer and the number of neurons in each layer are set in a timely manner. As will be described later with reference to FIG. 10, the neural network output is performed so that the shape control operation end of the Sendimia rolling mill, which is the output layer, outputs two types of output in the + direction and the-direction with respect to each control operation end. Consists of.

FIG. 8 shows the shape deviation and the control method. Here, the upper part of FIG. 8 shows the control method when the shape deviation is large, and the lower part of FIG. 8 shows the control method when the shape deviation is small. The height direction is the magnitude of the shape deviation, the horizontal axis direction is the plate width direction, and both sides of the plate width represent the plate end and the center represents the plate center. As shown in the upper part of FIG. 8, when the shape deviation is large, the correction of the overall shape is prioritized over the local shape deviation in the plate width direction. On the other hand, as shown in the lower part of FIG. 8, when the shape deviation is small, priority is given to reducing the local shape deviation.

In this way, since it is necessary to change the control method according to the magnitude of the shape deviation, the shape deviation step 202 is provided to the neural networks 101 and 111 as shown in FIG. 7, and the magnitude of the shape deviation is determined. .. Regarding the shape deviation, regardless of the magnitude of the shape deviation, it is preferable to use one standardized to 0 to 1, for example. This is just an example, and it is conceivable to input the shape deviation to the input layer of the neural network as it is without standardizing it, or to change the neural network itself according to the magnitude of the shape deviation (for example, two neural networks). It is also conceivable to prepare a net and separate the neural network used when the shape deviation is large and the neural network used when the shape deviation is small).

The neural networks 101 and 111 having the configuration shown in FIG. 7 described above are trained in the operation method for the shape pattern, and the shape control is performed using the trained neural network. Even neural networks with the same configuration have different characteristics depending on the learning conditions, and different control outputs can be output for the same shape pattern.

Therefore, by properly using a plurality of neural networks according to other conditions of the shape record, it is possible to configure the optimum control for various conditions. This is a response to specification B. The configuration of FIG. 4 described above shows a specific example in the case of making such a specification. In the configuration example of FIG. 4, the neural network 101 used in the control rule execution unit 10 is prepared as a separate neural network according to the rolling results, the name of the rolling mill operator, the steel type of the material to be rolled, the plate width, and the like, and the control rule database is prepared. Register in DB1. The neural network selection unit 102 selects a neural network that matches the conditions at that time, and sets it in the neural network 101 of the control rule execution unit 10. As a condition at that time in the neural network selection unit 102, it is preferable to take in the plate width data from the actual data Si in the controlled plant 1 and select the neural network accordingly. Further, if the plurality of neural networks used here have an input layer and an output layer as shown in FIG. 7, the number of layers of the intermediate layer and the number of units of each layer may be different.

FIG. 9 shows an outline of the control input data creation unit 2 that creates data S1 (normalized shape deviation 201, shape deviation step 202) for inputting to the input layers of the neural networks 101 and 111. Here, as the actual data Si, the shape detector data of the shape detector that detects the plate shape at the time of rolling in the Sendimia rolling mill, which is the controlled plant 1, is input, and first, each shape deviation PP value calculation device 210 is used. _{The shape deviation PP value (Peak To Peak value) S PP} , which is the difference between the maximum value and the minimum value of the detection result of the shape detector zone, is obtained. The shape deviation step calculation device 211 classifies the shape deviation into three stages of large, medium, and small according to the _{shape deviation PP value SPP.} The shape is the distribution of the elongation rate of the material to be rolled in the plate width direction, and I-UNIT, which expresses the elongation rate in 10-5 units, is used as a unit. For example, it is classified as follows.

Here, the shape deviation stage is (large = 1, medium = 0, small = 0) due to the establishment of equation (1), and the shape deviation stage is (large = 0, medium = 1, small) due to the establishment of equation (2). = 0), and the shape deviation stage is classified as (large = 0, medium = 0, small = 1) by the establishment of equation (3). Here, the shape deviation of each zone is standardized using S _{PM with S PM} = S _PP.

As described above, the normalized shape deviation 201 and the shape deviation step 202, which are the input data to the neural network 101, are created. The standardized shape deviation 201 and the shape deviation step 202 are the input data S1 of the control rule execution unit 10.

FIG. 10 shows an outline of the control output calculation unit 3. The control output calculation unit 3 is a control operation end operation command S2 (in the case of shape control of the Sendimia rolling mill, AS-U operation degree 301, first intermediate) which is an output from the neural network 101 in the control rule execution unit 10. From the operation degree 302 corresponds to this), the control operation amount S3 which is an operation command to each shape control operation end is created. Here, one data example is shown for each of the AS-U operation degree 301 and the first intermediate operation degree 302 in which a plurality of numbers exist, and each data is composed of a pair of data of the opening direction degree and the closing direction degree. Has been done.

In the control output calculation unit 3, the input AS-U operation degree 301 has outputs in the open direction and the closed direction of each AS-U. _Therefore, by multiplying the difference between them by the conversion gain G ASU, each AS- Outputs an operation command to U. Since the control output to each AS-U is the AS-U position change amount (unit is length), the conversion gain G _{ASU is the conversion gain from the degree to the position change amount.}

Further, since the input first intermediate operation degree 302 has outputs of the first intermediate outer side and the inner side, the operation command to each first intermediate roll shift is output by multiplying _{the difference between them by the conversion gain G 1ST.} do. Since the control output to each first intermediate roll is the first intermediate roll shift position change amount (unit is length), the conversion gain G _{1ST is the conversion gain from the degree to the position change amount.}

From the above, the control operation amount S3 can be calculated. The control operation amount S3 is composed of # 1 to # nAS-U position change amount (n depends on the number of saddles of the AS-U roll), the upper first intermediate shift position change amount, and the lower first intermediate shift position change amount. ing. Note that FIG. 10 shows a system for adding disturbance data from the control operation disturbance generation unit 16 to the control operation end operation command S2.

The outline of the operation of the evaluation function setting unit 17 will be described with reference to FIG. The evaluation function reflects the operator's intention to control the shape of the rolling mill. The operator's intention changes according to the rolling state. Here, it is assumed that the rolling state is distinguished by the rolling speed. As shown in FIG. 11, the rolling speed of the rolling mill changes from a stopped state to accelerating, rolling at a constant speed, decelerating, and stopping. The rolling state also changes to 17-1, 17-2, 17-3, ... According to the change in the rolling speed. Then, the operator's intention also changes according to the change in the rolling state, such as intention 1, intention 2, intention 3, and so on. The operator's intention is, for example, as follows.

<Intention 1> At the beginning of rolling at low speed, priority is given to the central part of the plate in order to ensure the stability of the plate.
<Intention 2> When accelerating rolling, the edge of the plate is emphasized in order to prevent meandering of the plate.
<Intention 3> When the rolling speed is constant, the quality of the material to be rolled is taken into consideration, and the shape deviation in the elongation direction of the plate end is allowed so that the plate breakage does not occur, and the shape of the central portion is prioritized.

The evaluation functions A1 to AN are associated with each of the above intentions as follows.
The evaluation function A2 corresponds to the intention 1.
The evaluation function A1 corresponds to the intention 2.
The evaluation function A3 corresponds to the intention 3.

The evaluation function DB DB 5 stores the correspondence between the operator's intention and the evaluation function as described above. FIG. 12 shows an example of the evaluation function DB DB5. It is defined which of the evaluation functions A1 to A6 (evaluation function NO) is used for each intention of the operator according to the above-mentioned rolling state.

Since the rolling states to which the intents 1, 2 and 3 are applied can be distinguished by the rolling speed, it is possible to select which of the evaluation functions A1 to AN is used according to the rolling speed. An operator, an operation engineer, or the like manually sets the association between the rolling speed and the evaluation functions A1 to AN in the evaluation function DB DB5 by using the evaluation function manual setting unit 171. The evaluation function selection unit 172 selects an evaluation function corresponding to the operator's intention according to the rolling state determined from the rolling performance Si (including the actual value of the rolling speed) according to the setting, and controls the control output determination unit 5 and the control. It is set in the result quality determination unit 6.

The manual setting of the evaluation function selection by the operator or the operation engineer is different from the actual one because the actual operator's judgment is not set properly or the operator discovers and uses a new judgment criterion. There can be cases. In order to evaluate the quality of this manual setting, the evaluation function selection method learning unit 173 determines the quality of the evaluation function selection method based on the actual data obtained in the actual rolling operation. Further, when it is determined that the selection method is not good, the evaluation function selection method learning unit 173 changes the setting of the evaluation function selection method in the evaluation function database DB5.

FIG. 13 is a diagram for explaining an outline of the operation of the evaluation function selection method learning unit 173. If the operator determines that the shape of the plate is bad during the rolling operation, the operator starts the manual operation, and continues the manual operation until it determines that the shape has improved. Therefore, the intention of the operator is reflected when the operator starts the manual operation and when the manual operation is completed. The evaluation function selection method learning unit 173 calculates the shape evaluation results of each evaluation function A1 to AN from the data at that time, and if the shape evaluation results are compared with each other, the relative quality of the evaluation function, that is, which It can be determined whether the evaluation function is close to the operator's intention.

Assuming that the smaller the value of the shape evaluation value, the better the shape, the evaluation function has a large shape evaluation value at the start of the manual operation and a small shape evaluation value at the end of the manual operation. , It can be judged that the evaluation function is suitable for the rolling state (rolling speed).

In this embodiment, the calculation method differs for each evaluation function, such as an evaluation function that uses the squared average and an evaluation function that uses the maximum or minimum value. Therefore, a common index is used to evaluate the quality of the evaluation function. It is necessary to compare by using it as an index (evaluation function quality index). Here, as an example, the evaluation function selection method learning unit 173 compares the evaluation functions using the ratio Xi shown in the following equation.
Ratio Xi = (ab) / b
a is a shape evaluation value at the time when the manual operation is started. b is a shape evaluation value at the time when the manual operation is completed. The evaluation function selection method learning unit 173 evaluates the evaluation function Ai in which the ratio Xi, which is the evaluation function quality index, has the largest value among the evaluation functions A1 to AN, which best suits the operator's intention in the rolled state at that time. Judge that it is an evaluation function that can be obtained, and select it as the best evaluation function.

The rolling state at the start or end of the manual operation and the operator's intention at that time can be determined from the rolling results. The evaluation function selection method learning unit 173 corresponds to the intention of the corresponding operator if the evaluation function associated with the intention of the corresponding operator in the evaluation function DB DB 5 is different from the best evaluation function selected here. Update the evaluation function to be evaluated to the evaluation function Ai. Then, from the next time, the evaluation function selection method learning unit 173 sets the evaluation function Ai in the control output determination unit 5 and the control result pass / fail determination unit 6 according to the changed settings.

The graph of FIG. 13 shows the time transition of the shape evaluation values of the two evaluation functions A1 and A2. The shape evaluation value at the time when the manual operation of the evaluation function A1 is started is A1S, and the shape evaluation value at the time when the manual operation is completed is A1E. The shape evaluation value at the time when the manual operation of the evaluation function A2 is started is A2S, and the shape evaluation value at the time when the manual operation is completed is A2E.

As is clear from FIG. 13, the ratio X2 = (A2S-A2E) / A2E of the evaluation function A2 is larger than the ratio X1 = (A1S-A1E) / A1E of the evaluation function A1.

Further, considering the possibility that the evaluation function itself set manually is not appropriate, the evaluation function learning unit 174 learns the evaluation function.

FIG. 14 is a diagram for explaining an outline of the operation of the evaluation function learning unit 174. The evaluation function learning unit 174 is for an evaluation function that outputs the shape evaluation value by inputting the shape result which is the actual value of the shape of the plate obtained by rolling and the rolling result which is the parameter value of the control operation in rolling. A neural network (neural network for evaluation function) is set up, and the neural network for the evaluation function is trained using the actual data. For the rolling performance to be input to the neural net for the evaluation function, it is preferable to select a rolling performance (for example, rolling speed) that is likely to affect the evaluation function. The trained neural network can be used as an evaluation function.

As described above, the evaluation of the shape intended by the operator can be interpreted that the shape of the plate is bad when the operator starts the manual operation and the shape of the plate is good when the manual operation is completed. Therefore, in the process of producing the plate by the rolling mill, the evaluation function learning unit 174 sets the shape evaluation value = 1 (1 indicates that the shape is bad) at the time when the operator starts the manual operation, and ends the manual operation. Assuming that the shape evaluation value at the time point = 0 (0 indicates that the shape is good), the shape and rolling results at that time are accumulated as teacher data. Then, the evaluation function learning unit 174 performs supervised learning of the neural network using the accumulated teacher data. As a result, the trained neural network can be used as an evaluation function because the shape evaluation value is output when the rolling result and the shape result are input.

FIG. 15 shows an outline configuration of the evaluation function learning unit 174. The control output determination unit 5 and the control result quality determination unit 6 initially use an evaluation function manually set by the operator. The evaluation function learning unit 174 is a neural net for an evaluation function that provides an evaluation function in place of the initial evaluation function by adding the teacher data described later to the rolling result data Si including the shape result and the rolling result and learning as learning data. To build.

The evaluation function execution unit 174 has an evaluation execution unit and a learning execution unit.

The evaluation execution unit includes an evaluation function neural network 1740 used in the control output determination unit 5 and the control result quality determination unit 6, and evaluates using the evaluation function neural network 1740.

The learning execution unit includes an evaluation function neural net 1741 equivalent to the evaluation function neural net 1740, and performs learning using the evaluation function neural net 1741. Here, as shown in FIG. 14, the evaluation function neural network 1741 is a neural network that inputs the shape result and the rolling result and outputs the shape evaluation value. At the time of learning the evaluation function neural net 1741, the rolling record data Si including the shape record and the rolling record is used as input data, the shape evaluation value described later is used as teacher data, and the combination thereof is used as training data. Therefore, if the combination of the shape result, the rolling result, and the teacher data is stored as the learning data in the evaluation function learning data database 1743, and the learning data is accumulated to a certain extent, the learning execution unit performs the learning of the neural net. good.

In addition to the evaluation function neural net 1741 described above, the learning execution unit includes an evaluation function neural net learning control unit 1744, an input data creation unit 1745, and a teacher data creation unit 1746.

The teacher data creation unit 1746 creates teacher data with a shape evaluation value = 1 at the timing when the manual operation is started by using the signal of the manual operation for the operator's shape. Further, the teacher data creation unit 1746 notifies the input data creation unit 1745 of the timing at which the manual operation is started. The input data creation unit 1745 acquires the shape record and the rolling record at the timing when the manual operation is started, and uses them as input data. The input data created by the input data creation unit 1745 and the teacher data created by the teacher data creation unit 1746 are stored in the evaluation function learning data database 1743 as a set of training data.

Similarly, the teacher data creation unit 1746 creates teacher data with a shape evaluation value of 0 at the timing when the manual operation is completed by using the signal of the manual operation for the operator's shape. Further, the teacher data creation unit 1746 notifies the input data creation unit 1745 of the timing when the manual operation is completed. The input data creation unit 1745 acquires the shape record and the rolling record at the timing when the manual operation is completed, and uses them as input data. The input data created by the input data creation unit 1745 and the teacher data created by the teacher data creation unit 1746 are stored in the evaluation function learning data database 1743 as a set of training data.

When the training data is accumulated to some extent (for example, 1000 sets) in the evaluation function learning data database 1743, the evaluation function neural net learning control unit 1744 reads the training data from the evaluation function learning data database 1743, and inputs the input data from the training data. The teacher data is acquired and given to the evaluation function neural net 1741, and the neural net is trained.

When the learning of the evaluation function neural net 1741 is completed in the learning execution unit, the evaluation function neural net 1741 is copied to the evaluation function neural net 1740 of the evaluation execution unit. As a result, the evaluation function neural network 1740 is updated with a new one. As a result, the control output determination unit 5 and the control result quality determination unit 6 can perform evaluation by a new neural network for the evaluation function.

In this embodiment, it is considered that the suitable evaluation function will be different if the conditions such as the plate width and plate thickness to be the control target and the steel type of the material are different. The neural net may be stored in the evaluation function database DB5 as an evaluation function, and they may be used properly according to the conditions. In addition, it is possible to cover with one neural network by considering the plate width, plate thickness, steel type, etc. as the rolling results.

Until the learning progresses to some extent, the numerical value of the evaluation function obtained from the neural network for the evaluation function may be inaccurate. Therefore, the evaluation function selection method learning unit 173 may select and use the evaluation function in consideration of not only the values of the setting evaluation functions A1 to AN but also the rolling state.

As described above, the evaluation function setting unit 17 sets the optimum evaluation function according to the rolling state in the control output setting unit 5 and the control result quality determination unit 6.

FIG. 16 is a diagram for explaining the outline of the control output determination unit 5. The control output determination unit 5 is composed of a rolling phenomenon model 501 and a shape correction pass / fail determination unit 502, and includes actual data Si from the controlled plant 1, control operation amount S3 from the control output calculation unit 3, and an output determination database. The information of DB3 is obtained, and the control operation amount output availability data S4 is given to the control operation end. With this configuration, in the control output determination unit 5, the change in shape when the control operation amount S3 calculated by the control output calculation unit 3 is output to the rolling mill, which is the control target plant 1, is reported to the known control target plant 1. Predict by inputting to the model (rolling phenomenon model 501 in the case of the embodiment of FIG. 16), and if the shape is expected to deteriorate, the control operation amount output SO is suppressed to prevent the shape from deteriorating significantly. do.

More specifically, the control operation amount S3 is input to the rolling phenomenon model 501, the shape change due to the control operation amount S3 is predicted, and the shape deviation correction amount prediction data 503 is calculated. On the other hand, the shape deviation prediction data 505 is obtained by adding the shape deviation correction amount prediction data 503 to the shape detector data Si (current shape deviation actual data 504) from the controlled plant 1, and the shape deviation prediction data 505 is obtained. By evaluating, it is possible to predict how the shape will change when the control operation amount S3 is output to the controlled plant 1. From the current shape deviation actual data 504 and shape deviation prediction data 505, the shape correction pass / fail judgment unit 502 determines whether the shape changes in the direction of improvement or deterioration, and the control operation amount output availability data. Obtain S4.

Specifically, the shape correction quality determination unit 502 determines the quality of the shape correction as follows. First, in order to consider the control priority in the plate width direction, the quality of the shape change is determined by using the evaluation function according to the rolling state set from the evaluation function setting unit 17. For example, the evaluation function J shown in the following equation is used to determine the quality of the shape change. In the following formulas, εfb (i) the shape deviation results 504, εest (i) the shape deviation estimation 505, i is the shape detector zone, rand is a random number _{section, J Ai} in the evaluation function set by the evaluation function setting unit 17 be.

When the evaluation function J of the above equation is used, the evaluation function J is positive when the shape is good, and the evaluation function J is negative when the shape is bad. Further, land is a random number term, and the evaluation result of the evaluation function J is randomly changed. As a result, even if the shape deteriorates, the evaluation function J may be positive. Therefore, even if the rolling phenomenon model 501 is incorrect, the relationship between the shape pattern and the control method should be learned. Is possible. Here, the random value should be increased in the maximum value when the model of the controlled plant 1 is uncertain as in the initial test run, and set to 0 when the control method is learned to some extent and stable control is to be performed. change.

In the shape correction pass / fail judgment unit 502, the evaluation function J is calculated, and when J ≧ 0, the control operation amount output availability data S4 = 1 (possible), and when J <0, the control operation amount output availability data S4 = 0 (possible). The control operation amount output availability data S4 is output as in (No).

The control output suppression unit 4 determines whether or not the control operation amount output SO is output to the controlled target plant 1 according to the control operation amount output availability data S4, which is the determination result of the control output determination unit 5. The control operation amount output availability data S4 is # 1 to # nAS-U position change amount output, upper first intermediate shift position change amount output, and lower first intermediate shift position change amount output.
IF (control operation amount output availability data S4 = 0) THEN
# 1 to # nAS-U Position change amount output = 0
Upper 1st intermediate shift position change amount output = 0
Lower 1st intermediate shift position change amount output = 0
ELSE
# 1 to # nAS-U position change amount output = # 1 to # nAS-U position change amount Upper 1st intermediate shift position change amount output = Upper 1st intermediate shift position change amount Lower 1st intermediate shift position change amount output = Lower 1st intermediate shift position change amount ENDIF
Is determined by.

In the control execution device 20, shape control is performed by executing the above calculation from the actual data Si from the controlled target plant 1 (rolling machine) and outputting the control operation amount output SO to the controlled target plant 1 (rolling machine). To carry out.

Next, an outline of the operation of the control method learning device 21 will be described. In the control method learning device 21, the time delay data of the data used in the control execution device 20 is used. Time delay ^{Z -1 means} the ^{e -TS,} indicating that delaying preset time T. Since the controlled plant 1 has a time response, there is a time delay until the actual data changes due to the control operation amount output SO. Therefore, the learning is performed using the actual data at the time when the delay time T has elapsed after the control operation is executed. In shape control, it takes several seconds for the shape meter to detect the shape change after the operation command is output to the AS-U and the first intermediate roll, so it is better to set T = 2 to 3 seconds (of the shape detector). Since the delay time changes depending on the type and rolling speed, the optimum time until the change of the control operation end becomes the shape change may be set as T).

FIG. 17 is a diagram for explaining an outline of the operation of the control quality determination unit 6. In the shape change pass / fail judgment unit 602, the _{pass / fail judgment evaluation function JC} as shown in the following equation is used.

In the above formula, εfb (i) is the actual shape deviation actual data contained in the data Si, εlast (i) the shape deviation actual data preceding value, the J _Ai is the evaluation function set by the evaluation function setting unit .. Here, the evaluation function J _Ai, the evaluation function J _Ai manually set in advance in the evaluation function setting unit 17 or the evaluation function learning unit _174, sets the evaluation learned function (learning evaluation function). The quality judgment evaluation function Jc determines the quality of the control result. Also, when the control operation amount output availability data S4, which is the determination result of the control output determination unit 5, is 0 (control output is not possible), the control operation amount output to the controlled target plant 1 is actually 0, but the shape is Judge that it has become worse.

Here, when the control operation amount output availability data S4 = 0, the control result quality data S6 = -1 is set. Further, the threshold upper limit LCU and the threshold addition / subtraction LCL are set in advance under the threshold condition (LCU ≧ 0 ≧ LCL). At this time, if the result of comparison with the pass / fail judgment evaluation function Jc is Jc> LCU, the control result pass / fail data S6 = -1 (the shape has deteriorated) is set.
If LCU ≧ Jc ≧ 0, the control result good / bad data S6 = 0 (changes in the direction of worsening the shape) is set.
If 0> Jc ≧ LCL, the control result quality data S6 = 1 (changes in the direction of improving the shape) is set.
If Jc <LCL, it is assumed that the control result quality data S6 = 0 (the shape has improved).

Here, since the shape of the control result good / bad data S6 = -1 has deteriorated, when the output control output is suppressed, the control result good / bad data S6 = 0 is output because there is no shape change or the shape has improved. When the control output is held, the control result quality data S6 = 1 has changed in the direction of improving the shape, but there is a possibility that the shape will be further improved. Therefore, this is a case where the output control amount is increased.

If the evaluation function _JAi is different, the quality judgment evaluation function Jc is different. Therefore, it is conceivable that the determination result of the control result quality data S6 is also different. Therefore, in the control method learning device 21, the determination of the control result quality data S6 is performed for each preset evaluation function.

Next, the outline of the learning data creation unit 7 will be described. As shown in FIG. 3, in the learning data creation unit 7, the control operation end operation command S2, the control operation amount S3, based on the determination result (control result quality data S6) from the control result quality determination unit 6. From the determination result of the control output suppression unit (control operation amount output availability data S4), the teacher data S7a for the neural network 111 used by the control rule learning unit 11 is created.

The teacher data S7a in this case has an AS-U operation degree 301 and an intermediate operation degree 302, which are outputs from the output layer of the neural network 111, as shown in FIG. The learning data creation unit 7 includes a control operation end operation command S2 (AS-U operation degree 301, 1 intermediate operation degree 301) which is an output of the neural network 101, and # 1 to # nAS-U which is a control operation amount output SO. The teacher data S7a for the neural network 111 used in the control rule learning unit 11 is created by using the position change amount output, the upper first intermediate shift position change amount output, and the lower first intermediate shift position change amount output.

In explaining the outline of the operation of the learning data creation unit 7, the relationship between the data and symbols of each unit in the control output calculation unit 3 of FIG. 10 is organized in FIG. Here, the AS-U operation degree 301 is typically shown for the control operation end operation command S2 which is the output of the neural network 101, the data on the operation degree positive side is OPref, the data on the operation degree negative side is OMref, and control. The operation degree generated randomly from the operation disturbance generation unit 16 will be described as an operation degree random number Olef, the conversion gain will be G, and the control operation amount output SO will be Clef. As described above, here, for the sake of simplicity, the output from the output layer of the neural network 101 of the control rule execution unit 10 is randomly generated from the operation degree positive side and the operation degree negative side, and the control operation disturbance generation unit 16. The degree of operation to be performed is a random number of degree of operation. Further, the control operation amount output SO for the control operation end is set as the operation command value.

FIG. 19 shows a processing stage and processing contents in the learning data creation unit 7. Here, to explain according to the promise of the symbols in FIG. 18, in the first processing step 71, the operation command value Clef is obtained by the equation (6).

In the next processing step 72, the operation command value Clef is modified to be C'ref according to the control result quality data S6. Specifically, the operation command is based on the equation (7) when the control result quality data S6 = -1, the equation (8) when the control result quality data S6 = 0, and the equation (9) when the control result quality data S6 = 1. Let the modified value C'ref of the value Clef.

In the processing step 73, the operation degree correction amount ΔCref is obtained from the corrected operation command value C'ref by the equations (10) and (11).

In the processing step 74, the teacher data OP'ref and OM'ref to the neural network 111 are obtained by the equation (12).

As described above, in the learning data creation unit 7, as shown in FIG. 18, the operation command value Clef actually output to the controlled target plant 1 is the control result good / bad data S6 which is the judgment result in the control result good / bad judgment unit 6. The operation command value correction value C'ref is calculated according to the above. Specifically, when the control result quality data S6 = 1, the control direction is OK, but when it is determined that the control output is insufficient, the operation command value is increased by ΔClef in the same direction. To. On the contrary, when the control result quality data S6 = -1, the operation command value is reduced by ΔClef in the opposite direction when it is determined that the control direction is wrong. Since the conversion gain G is known because it is set in advance, it is possible to obtain the correction amount ΔOref if the values on the positive operation degree side and the negative operation degree side are known. Here, ΔClef is set by obtaining an appropriate value in advance by simulation or the like. By the above procedure, the teacher data OP'ref and OM'ref used in the control rule learning unit 11 can be obtained by the above equation (12).

Although the description is given with a simple example in FIG. 19, in reality, the AS-U operation degree 301 for # 1 to # nAS-U, the upper first intermediate roll shift, and the lower first intermediate roll shift are the first. All of the 1 intermediate operation degree 302 is carried out and used as the teacher data (AS-U operation degree teacher data, 1 intermediate operation degree teacher data) of the neural network 111 used in the control rule learning unit 11.

FIG. 20 shows an example of data stored in the learning data database DB2. In order to learn the neural network 111, a combination of a large number of input data S8a and teacher data S7a is required. Therefore, the teacher data S7a (AS-U operation degree teacher data, first intermediate operation degree) created by the learning data creation unit 7 is the input data S1 (standard) input to the control rule execution unit 10 by the control execution device 20. It is stored in the training data database DB2 as a set of training data S11 in combination with the time delay data S8a of the formed shape deviation 201 and the shape deviation stage).

In the plant control device of FIG. 3, various databases DB1, DB2, DB3, DB4, and DB5 are used, but in FIG. 20, each database DB1, DB2, DB3, and DB4 are managed and operated in an interconnected manner. The configuration of the neural net management table TB of is shown. The management table TB includes a specification management table. Specifically, the management table TB is classified according to (B1) plate width, (B2) steel grade, and evaluation functions A1 to AN regarding control priority. (B1) As the plate width, for example, 4 divisions of 3 feet width, meter width, 4 feet width, and 5 feet width are used, and as the steel grade, about 10 divisions of steel grades (1) to (10) are used. Further, the control evaluation function is of N (N is the number of set evaluation functions. In this embodiment, N = 6). In this case, there are 80 divisions, and 240 neural networks are used properly according to the rolling conditions.

The neural network learning control unit 112 applies the learning data, which is a combination of the input data and the teacher data, as shown in FIG. 20, according to the neural network management table TB of FIG. Is stored in the learning data database DB2 as shown in FIG. 22.

Every time the control execution device 20 executes shape control on the controlled plant 1, N sets of learning data are created according to the evaluation function. This is because N types of teacher data are created because the control result pass / fail judgment is performed using N evaluation functions for the control priority for the same input data and control output. When the teacher data is accumulated to some extent (for example, 200 sets) or newly accumulated in the learning data database DB2, the neural network learning control unit 112 instructs the learning of the neural network 111.

A plurality of neural networks are stored in the control rule database DB1 according to the management table TB as shown in FIG. 21, and in the neural network learning control unit 112, the neural network Nos. Is specified, the neural network selection unit 113 extracts the neural network from the control rule database DB1 and sets it in the neural network 111. The neural network learning control unit 112 instructs the input data creation unit 114 and the teacher data creation unit 115 to take out the input data and the teacher data corresponding to the neural network from the learning data database DB2, and uses them to make a neural network. The learning of the net 111 is carried out. Various methods have been proposed for learning the neural network, and any method may be used.

When the learning of the neural network 111 is completed, the neural network learning control unit 112 sends the neural network 111, which is the learning result, to the neural network No. of the control rule database DB1. Learning is completed by writing back to the position of.

The learning may be performed simultaneously for all the neural networks defined in FIG. 21 at regular time intervals (for example, every day), or the neural network in which new learning data is accumulated to some extent (for example, 100 sets). No. Only the neural network of the above may be trained at that time.

As a result, the shape of the rolling mill, which is the controlled plant 1, is not significantly disturbed.
1) Rather than setting the reference shape pattern and the control operation for it separately in advance and learning the control operation method, the combination of the shape pattern and the control operation is learned and the control operation is performed using it. ..
2) New control rules are not predictable in advance, and control rules that could not be predicted at all may be optimal. Therefore, operate the control operation end at random and find it while looking at the control results for it. go.
3) The evaluation function for determining the control priority for the controlled object is set so as to match the operator's feeling and the operator's manual operation method according to the state of the controlled object.
Things can be realized.

The control rule database DB1 stores the neural network used by the control execution device 20, but if the stored neural network is only the initial processing performed with random numbers, the learning of the neural network proceeds. However, it takes time until some control becomes possible. Therefore, when the control unit is constructed for the controlled target plant 1, the control rules are learned in advance by simulation based on the control model of the controlled target plant 1 known at that time, and the simulator is used. By storing the learned neural network in the database, it is possible to control the performance to some extent from the beginning of the controlled plant.

In addition, since the neural network is learned using an evaluation function that matches the manual operation method of the operator, the operator does not have to manually operate the change of the control target due to the control output, which reduces the load on the operator. It is also possible to improve control accuracy and operation efficiency.

The embodiments described above include the following items. However, the matters included in the embodiment are not limited to the matters shown below.

The control device of the present disclosure is a control device that controls a control target, and is a control execution device that gives a control output to the control target according to a given control rule, and the control target using a designated evaluation function. A control method in which the given control output is evaluated, training data is created using the evaluation result, the control rule is constructed by learning the training data, and the control rule is given to the control execution device. A learning device and a plurality of evaluation functions are held in advance, one of the plurality of evaluation functions is selected based on the control state of the controlled object, and the selected evaluation function is used as the control method learning device. It has an evaluation function setting unit specified in.

According to this configuration, the control output is given to the control target according to the control rule constructed by learning the learning data using the evaluation result of the evaluation for the control output by the evaluation function selected based on the control state, so that the control result is appropriate. It is expected that control based on good quality judgment will be feasible.

Further, according to the present disclosure, the evaluation function setting unit calculates an evaluation function quality index for each of the plurality of evaluation functions based on the control state of the controlled object and the manual operation by the operator, and the evaluation function. The evaluation function specified in the control method learning device is selected based on the pass / fail judgment index. According to this configuration, by utilizing the relationship between the manual operation by the operator and the control state of the controlled object, it becomes easy to select an evaluation function in which the control intended by the operator is highly evaluated.

Further, according to the present disclosure, the evaluation function setting unit calculates the evaluation value of the evaluation function at the timing when the operator starts the manual operation and the timing when the operator finishes the manual operation, and the evaluation value. Is used to calculate the evaluation function pass / fail judgment index. According to this configuration, the operator starts the manual operation when it is judged that the shape of the plate is bad during the rolling operation, and continues the manual operation until it is judged that the shape is improved. Therefore, the operator is operated from the evaluation value at that time. Can get the intention of.

Further, according to the present disclosure, the evaluation function setting unit has an evaluation value a of the evaluation function at the timing when the operator starts the manual operation and an evaluation value a of the evaluation function at the timing when the operator finishes the manual operation. b is calculated, and the evaluation function pass / fail judgment index is calculated as (ab) / b. According to this configuration, even if the calculation method is different for each of a plurality of evaluation functions, it is possible to compare the evaluation function quality indexes with each other.

Further, according to the present disclosure, the evaluation function inputs the control output to the control target and the actual data of the control target reflecting the control result of the control output, and outputs the evaluation result. The evaluation function setting unit constructs the evaluation function by learning the learning data based on the manual operation by the operator, the control output to the control target, and the actual data of the control target. According to this configuration, since the manual operation of the operator is used, it is possible to construct an evaluation function that reflects the intention of the operator.

Further, according to the present disclosure, the evaluation function setting unit has the control output to the control target and the control target at the timing when the operator starts the manual operation and the timing when the operator finishes the manual operation. The evaluation function is constructed by learning the learning data based on the actual data. According to this configuration, the operator starts the manual operation when the shape of the plate is evaluated to be bad during the rolling operation, and continues the manual operation until the shape is evaluated to be good. Therefore, the evaluation value reflecting the evaluation of the operator is reflected. Can be used as training data to construct an evaluation function that makes an evaluation close to the operator's evaluation.

Further, according to the present disclosure, the evaluation function setting unit generates learning data with the evaluation value at the timing when the operator starts the manual operation as a predetermined value c, and the evaluation value at the timing when the operator finishes the manual operation. Is set as a predetermined value d, training data is generated, and the evaluation function is constructed by learning the training data. According to this configuration, the operator starts the manual operation when it is judged that the shape of the plate is bad during the rolling operation, and continues the manual operation until it is judged that the shape is improved. Therefore, the operator is operated from the evaluation value at that time. Can get the intention of.

Further, according to the present disclosure, the control execution device uses the control rule execution unit that gives control output to the control target according to the combination of the actual data of the control target and the control operation, and the control using the evaluation function. The control output determination unit that determines whether or not the control output output by the rule execution unit is applicable, and notifies the control method learning device that the combination of the actual data and the control operation is inappropriate if it is determined that the control output is not applicable. The control output determination unit includes a control output suppression unit that prevents the control output from being output to the control target when the control output determination unit determines that the application is not applicable. When the control output is actually output to the control target, the performance is used by using the evaluation function set by the evaluation function setting unit after a time delay until the control output is reflected in the performance data of the control target. Using the control result quality determination unit for determining whether the data is improved or deteriorated by the control output, the control result quality determination unit determined by the control result quality determination unit, and the control output. A learning data creation unit for obtaining teacher data and a control rule learning unit for learning the actual data and the teacher data as learning data are provided, and the control method learning device learns the data according to the state of the controlled plant. A separate combination of actual data and control operation was obtained for a plurality of control targets, and the combination of the obtained actual data and control operation was defined as the actual data and control operation of the controlled plant in the control rule execution unit. Used as a combination.

Further, the plant control device of the present disclosure is actually realized as a computer system, but in this case, a plurality of program groups are formed in the computer system.

These programs are, for example,
A control rule execution program that gives control output according to a defined combination of control target plant performance data and control operations to achieve the processing of the control execution device, and a judgment as to whether or not the control output output by the control rule execution program is possible. , When the control output judgment program and the control output judgment program that notify the control method learning device that the actual data and the control operation are incorrect, the control output is output to the controlled target plant, the actual data of the controlled target plant. It is a control output suppression program that prevents the control output from being output to the controlled plant when it is determined that
When the control execution device actually outputs the control output to the controlled plant in order to achieve the processing of the control method learning device, the actual data is displayed before the control after a time delay until the control effect appears in the actual data. A control result quality determination program for achieving the control result quality determination process for determining whether the control result is good or bad, and the control result quality in the control result quality determination program. It is a learning data creation program that obtains teacher data using control output, and a control rule learning program that learns the actual data and the teacher data as training data.
Then, by learning by the control method learning device, different combinations of actual data and control operations are obtained for a plurality of control targets according to the state of the controlled plant, and the obtained actual data and control operations are combined. Is used as a defined combination of the actual data of the controlled plant and the control operation in the control rule execution program.

In applying the apparatus of the present invention to an actual plant, it is necessary to determine the initial value of the neural network. In this regard, the combination of the actual data and the control operation is controlled before the controlled plant is controlled. It is preferable to use the control model of the controlled plant and create it by simulation to shorten the learning period of the combination of the actual data and the control operation in the controlled plant.

The present invention relates to, for example, a control method and a part of a rolling mill, which is one of rolling equipment, and there is no particular problem in actual application.

1: Control target plant 2: Control input data creation unit 3: Control output calculation unit 4: Control output suppression unit 5: Control output judgment unit 6: Control result quality judgment unit, 7: Learning data creation unit, 10: Control rule execution unit, 11: Control rule learning unit, 20: Control execution device, 21: Control method learning device, DB1: Control rule database, DB2: Output judgment database, DB3: Learning data database, Si: Actual data, SO: Control operation amount output, S1: Input data, S2: Control operation end operation command, S3: Control operation amount, S4: Control operation amount output availability data, S5: Good / bad judgment data, S6: Control result good / bad data, S7a, S7b, S7c: Teacher data, S8a, S8b, S8c: Input data (for control rule learning unit)

Claims (9)

A control device that controls a controlled object
A control execution device that gives control output to the control target according to a given control rule,
The control output given to the controlled object is evaluated using the specified evaluation function, learning data is created using the evaluation result, and the control rule is constructed by learning the learning data. , A control method learning device that gives the control rule to the control execution device,
An evaluation that holds a plurality of evaluation functions in advance, selects one of the plurality of evaluation functions based on the control state of the controlled object, and designates the selected evaluation function in the control method learning device. Function setting part and
Control device with. The evaluation function setting unit calculates an evaluation function good / bad index for each of the plurality of evaluation functions based on the control state of the controlled object and the manual operation by the operator, and based on the evaluation function good / bad judgment index, the evaluation function setting unit said. Control method Select the evaluation function to be specified for the learning device.
The control device according to claim 1. The evaluation function setting unit calculates the evaluation value of the evaluation function at the timing when the operator starts the manual operation and the timing when the operator finishes the manual operation, and uses the evaluation value to determine the quality of the evaluation function. Calculate the index,
The control device according to claim 2. The evaluation function setting unit calculates the evaluation value a of the evaluation function at the timing when the operator starts the manual operation and the evaluation value b of the evaluation function at the timing when the operator finishes the manual operation, and calculates the evaluation function b. Calculate the quality judgment index as (ab) / b,
The control device according to claim 3. The evaluation function inputs the control output to the control target and the actual data of the control target reflecting the control result of the control output, and outputs the evaluation result.
The evaluation function setting unit constructs the evaluation function by learning the learning data based on the manual operation by the operator, the control output to the control target, and the actual data of the control target.
The control device according to claim 1. The evaluation function setting unit obtains learning data based on the control output to the control target and the actual data of the control target at the timing when the operator starts the manual operation and the timing when the operator finishes the manual operation. By learning, the evaluation function is constructed.
The control device according to claim 5. The evaluation function setting unit generates learning data with the evaluation value at the timing when the operator starts the manual operation as a predetermined value c, and generates the learning data with the evaluation value at the timing when the operator finishes the manual operation as the predetermined value d. The evaluation function is constructed by generating and learning the training data.
The control device according to claim 6. The control execution device includes a control rule execution unit that gives control output to the control target according to a combination of actual data of the control target and a control operation, and the control output by the control rule execution unit using the evaluation function. The control output determination unit and the control output determination unit notify the control method learning device that the combination of the actual data and the control operation is inappropriate when the output is determined to be applicable or not. A control output suppression unit that prevents the control output from being output to the control target when it is determined that the control output is not applicable is provided.
The control method learning device is the evaluation function after a time delay until the control output is reflected in the actual data of the control target when the control execution device actually outputs the control output to the control target. Using the evaluation function set by the setting unit, the control result quality determination unit that determines whether the performance data is improved or deteriorated by the control output is determined by the control result quality determination unit and the control result quality determination unit. The control method learning device includes a learning data creation unit that obtains teacher data using the control result and the control output, and a control rule learning unit that learns the actual data and the teacher data as learning data. By learning, separate combinations of actual data and control operations are obtained for a plurality of control targets according to the state of the controlled plant, and the obtained combinations of actual data and control operations are combined in the control rule execution unit. Used as a defined combination of controlled plant performance data and control operations,
The control device according to claim 1. A control method for controlling a controlled object,
A control output is given to the controlled object according to the given control rule,
The control output given to the controlled object is evaluated using the specified evaluation function, and the control output is evaluated.
Create learning data using the evaluation results and
By learning the learning data, the control rule is constructed.
Select and specify one of a plurality of evaluation functions held in advance based on the control state for the control target.
A control method in which a computer performs that.

Images