JP2024034510A - Operation control device and method - Google Patents

Abstract

The present invention proposes a control device and method that can perform control closer to the sensibilities of an expert. [Solution] A control target for controlling the controlled object to a predetermined state is calculated, the calculated control target is transmitted to the controlled object, the controlled object is controlled, and the control target is controlled by manual intervention on the controlled object. 1 correction amount and the control result of the controlled object when the intervention operation is performed, and based on the obtained first correction amount and the control result of the controlled object, the first correction amount for the control result of the controlled object is obtained. learns the correction amount of the control target, obtains the control result of the current controlled object, and makes a second correction to the control target using the obtained control result of the controlled object and the learned model obtained by learning by the learning section. The control target is corrected using the calculated second correction amount, or the calculated second correction amount is transmitted to the controlled object. [Selection diagram] Figure 13

Description

The present invention relates to an operation control device and method, and is suitable for application to, for example, an operation control device that controls the operation of a furnace for firing carbon electrodes.

In recent years, the industry as a whole has been promoting the use of AI (Artificial Intelligence) for big data. In the manufacturing industry, operational process data acquired by sensors, etc., quality data acquired through sampling inspections, etc. are mathematically analyzed to determine the operating status of controlled equipment such as production lines with the aim of stabilizing quality and yield. There is a growing need for technology to optimize settings and settings.

In addition, as a technology for optimizing the operating status and set values of a controlled device, for example, Patent Document 1 describes an intervention operation that was previously performed by skilled or experienced operators based on their own judgment in a waste incinerator. , it is disclosed that an operation is executed in place of a skilled operator based on a learned model generated by machine learning.

JP 2021-8991 Publication

However, although Patent Document 1 proposes a method of automating the control of equipment and equipment by using machine learning and other statistical methods, automatic control and human intervention when there is human intervention are proposed. There is no disclosure regarding a control method that coordinates the intervention operations.

The present invention has been made in consideration of the above points, and attempts to propose a control device and method that can perform control that coordinates automatic control and human intervention, and that can perform control that is closer to the sensibilities of an expert. That is.

In order to solve this problem, in the present invention, in a control device that controls a controlled object to a predetermined state, a control target for controlling the controlled object to the predetermined state is calculated, and the calculated control target is applied to the controlled object. a control target calculation unit that controls the controlled object, a first correction amount due to a manual intervention operation on the controlled object, and control of the controlled object when the intervention operation is performed; a learning unit that acquires the result and learns the first correction amount for the control result of the controlled object based on the obtained first correction amount and the control result of the controlled object; and the current controlled object. a correction information calculation unit that calculates a second correction amount for the control target using the acquired control result of the controlled object and the learned model obtained by learning by the learning unit; and the correction information calculation unit corrects the control target using the calculated second correction amount, or transmits the calculated second correction amount to the controlled object.

The present invention also provides a control method executed by a control device that controls a controlled object to a predetermined state, the method comprising: calculating a control target for controlling the controlled object to the predetermined state; A first step of controlling the controlled object by transmitting it to the controlled object, a first correction amount by a manual intervention operation on the controlled object, and the control when the intervention operation is performed. a second step of learning the first correction amount for the control result of the controlled object based on the obtained first correction amount and the control result of the controlled object; Acquire the current control result of the control target, and calculate a second correction amount for the control target using the acquired control result of the control target and the learned model obtained by learning by the learning unit. and in the third step, the control device corrects the control target using the calculated second correction amount, or corrects the control target using the calculated second correction amount. Now it will be sent to the target.

According to the control device and method of the present invention, it is possible to perform control in which automatic control and human intervention are coordinated.

According to the present invention, it is possible to realize a control device and method that can perform control closer to the sensibility of an expert.

FIG. 1 is a block diagram showing the overall configuration of an operation control system according to the present embodiment. FIG. 2 is a block diagram showing the logical configuration of the present operation control system. (A) to (E) are block diagrams for explaining the functions of each functional unit. It is a graph used to explain a temperature profile. It is a chart showing the structure of a manufacturing information table. It is a chart showing the structure of a pre-process information table. It is a chart showing the structure of an operation information table. It is a chart showing the composition of a control history table. It is a chart showing the composition of a control target value table. It is a chart showing the composition of an automatic correction value table. 3 is a diagram showing the structure of a device status information table. It is a chart showing the composition of an alert information table. FIG. 2 is a block diagram showing a series of operation control flows for a controlled device in the operation control system. FIG. 3 is a diagram illustrating an example of a screen configuration of a controlled device status display screen. It is a flowchart which shows the processing procedure of operation control processing. It is a flowchart which shows the processing procedure of relearning processing. 3 is a flowchart showing the processing procedure of screen creation and display processing.

An embodiment of the present invention will be described in detail below with reference to the drawings.

(1) Configuration of the operation control system according to the present embodiment In FIG. 1, reference numeral 1 indicates the operation control system according to the present embodiment as a whole. The operation control system 1 includes a controlled device 3 and an operation control device 4 connected via a network 2. Note that the controlled device 3 may be any type of equipment or device, but in the following, it is assumed that it is a furnace for firing a carbon electrode.

The operation control device 4 is a computer device having a function of controlling the operation of the controlled device 3 (hereinafter referred to as control of the furnace temperature in the controlled device 3), and includes a CPU (Central Processing Unit) 10, a main memory The device includes a device 11, a secondary storage device 12, an input device 13, and a display device 14.

The CPU 10 is a processor that controls the operation of the entire operation control device 4. Further, the main storage device 11 is composed of, for example, a semiconductor memory, and is used as a work memory of the CPU 10. In the case of this embodiment, a manufacturing information table 21, a pre-process information table 22, a physical/empirical model 23, an operation information table 24, and a control history table 25 constitute input data 20, which will be described later, and output data 30, which will be described later. An automatic correction value table 32, a control target value table 31, an alert information table 34, an apparatus status information table 33, and a learned model 40, which will be described later, are stored and retained in the main storage device 11.

The secondary storage device 12 is composed of a large-capacity nonvolatile storage device such as a hard disk device or an SSD (Solid State Drive), and stores various programs and data that need to be stored for a long period of time. The program stored in the secondary storage device 12 is loaded into the main storage device 11 when the operation control device 4 is started or when necessary, and the CPU 10 executes the program loaded into the main storage device 11 to perform operations as described below. Various processes are executed by the operation control device 4 as a whole. A control target value calculation program 50, a correction information calculation program 51, a learning program 52, and a display information creation program 53, which will be described later, are also stored and retained in this secondary storage device 12.

The input device 13 is used by the user to input necessary information and instructions to the operation control device 4, and includes, for example, a keyboard and a mouse. Further, the display device 14 is used to display necessary information, and is composed of, for example, a liquid crystal panel or an organic EL (Electro-Luminescence) panel. Note that the input device 13 and the display device 14 may be configured by an integrated touch panel.

FIG. 2 shows the logical configuration of the present operation control system 1. As shown in FIG. 2, the controlled device 3 includes an operation section 60, a control section 61, a valve drive section 62, and a detection section 63. The operation control device 4 also includes a control target value calculation section 64, a learning section 65, a correction information calculation section 66, a display information creation section 67, a manufacturing information table 21 that constitutes the input data 20, a pre-process information table 22, a physical/ The empirical model 23, the operation information table 24, the control history table 25, the control target value table 31, the automatic correction value table 32, the device status information table 33, the alert information table 34, and the learned model 40, which constitute the output data 30. It is composed of:

The operation unit 60 of the controlled device 3 allows an operator to manually adjust the valve opening degree of a gas valve (not shown) when the control of the controlled device 3 performed by the operation control device 4 is insufficient. It is a functional unit used to perform operations, and is composed of, for example, levers and operation panels. The operation unit 60 controls a correction value (a value corresponding to the valve opening degree of the gas valve, hereinafter referred to as a manual correction value) according to the content of the intervention operation performed by the worker. 61 and the operation control device 4, respectively. Note that this manual correction value is stored and managed in the control history table 25 on the operation control device 4 side.

The control unit 61 controls the gas flow notified from the operation control device 4 at predetermined timings (hereinafter referred to as control target value application timings) that arrive at a first period (for example, a period of 1 to several minutes) as described later. This is a functional unit that has the function of controlling the temperature inside the furnace by transmitting the target value of the valve opening degree of the valve (hereinafter referred to as the control target value) to the valve driving unit 62 that opens and closes the gas valve. . At this time, when a manual correction value is notified from the operation unit 60 or when an automatic correction value, which will be described later, is given from the operation control device 4, the control unit 61 changes the control target value to the manual correction value or the automatic correction value. The value is corrected, and the valve drive unit 62 is drive-controlled based on the corrected control target value.

The detection unit 63 detects various values related to the surrounding environment, such as the furnace temperature of the controlled device 3 controlled by the control unit 61 (hereinafter referred to as an operating value), and the temperature and humidity around the controlled device 3. It is a functional unit that detects a state (hereinafter referred to as an environmental condition as appropriate), and is composed of, for example, a plurality of sensors. The detection unit 63 uses the detected operating values and environmental conditions of the controlled device 3 as operating information 24X (FIG. 3(C)) to control the operation at a second cycle (for example, a 10-second cycle) that is shorter than the first cycle. Notify device 4. This operating information 24X is stored and managed in an operating information table 24, which will be described later, on the operation control device 4 side.

The control target value calculation unit 64 of the operation control device 4 executes the control target value calculation program 50 (FIG. 1) loaded into the main storage device 11 (FIG. 1) by the CPU 10 (FIG. 1) of the operation control device 4. This is a functional unit realized by. As shown in FIG. 3A, the control target value calculation unit 64 uses various information (hereinafter referred to as manufacturing information) 21X regarding the product to be manufactured stored in the manufacturing information table 21 in advance, and The operating value, which is the controlled value of the controlled device 3, is calculated based on various information related to the previous process stored in the information table 22 (hereinafter referred to as previous process information) 22X and the physical/empirical model 23 given in advance. (furnace temperature), and store the control target value 31X, which is the target value of the valve opening degree of the gas valve, in the control target value table 31 for each control target value application timing based on the calculated temperature profile. .

Note that the "physical/empirical model" here refers to an algorithm generated based on a heat transfer model or an empirical model (past performance of opening the gas valve to this extent when the temperature inside the furnace was around this extent). It is. Moreover, the "temperature profile" here refers to the operation plan from the start of operation to the end of operation of the controlled device 3 as shown in FIG. 4, expressed as a time series control target value 31X for each control target value application timing. refers to the data.

Further, the control target value calculation unit 64 calculates the operating value ( Furnace temperature) is controlled by feedforward or feedback control. Specifically, the control target value calculation unit 64 reads the corresponding control target value 31X stored in the control target value table 31 at each control target value application timing, and applies the read control target value 31X to the control target device 3. It is changed as necessary based on the operating value (furnace temperature) and then transmitted to the controlled device 3.

The learning unit 65 is a functional unit that is realized when the CPU 10 of the operation control device 4 executes the learning program 52 (FIG. 1) loaded into the main storage device 11. As shown in FIG. 3(B), the learning unit 65 uses the previous process information 22X stored in the previous process information table 22 and the past control contents for the controlled device 3 stored in the control history table 25, which will be described later. Based on the information (hereinafter referred to as control history information) 25 It learns whether the operation has been performed and generates and updates a trained model 40 made of a neural network, a nonlinear regression model (kernel model), a linear regression model, or the like.

In this case, the learning unit 65 performs the operator's intervention operation based on the amount of change in the operating value of the controlled device 3 in response to the operator's intervention operation (such as the first derivative value of the operating value or the difference value from the previous value). The learned model 40 may be generated and updated by learning the relationship between the correction amount and the amount of change in the operating value. By doing so, the technology of this embodiment can also be utilized for controlling thermal shock of the product (carbon electrode in this example) fired in the controlled device 3.

The correction information calculation unit 66 is a functional unit that is realized when the CPU 10 of the operation control device 4 executes the correction information calculation program 51 (FIG. 1) loaded into the main storage device 11. As shown in FIG. 3C, when the operating information 24X is transmitted from the detecting unit 63 of the controlled device 3, the correction information calculation unit 66 stores the operating information 24X in the operating information table 24. Further, the correction information calculation unit 66 inputs the operating information 24X and the previous process information 22X stored in the previous process information table 22 into the learned model 40 every time the operating information 24X is transmitted. , the automatic correction value 32X is calculated as a correction value that is the same as the manual correction value based on the intervention operation that is predicted to be performed by the worker under the current operating value of the controlled device 3 and the environmental conditions. However, the correction information calculation unit 66 may calculate the automatic correction value 32X using the pre-process information 22X or the like so that the control target value 31X of the controlled device 3 falls within a predetermined range. Then, the correction information calculation unit 66 transmits the calculated automatic correction value 32X to the controlled device 3 and stores it in the control history table 25 and the automatic correction value table 32, respectively.

In addition, the correction information calculation unit 66 monitors the presence or absence and magnitude of a manual correction value notified from the control unit 61 of the controlled device 3 as described above when an operator performs an intervention operation in the controlled device 3. do. If the manual correction value exceeds a preset range (hereinafter referred to as manual correction value tolerance range), the correction information calculation unit 66 temporarily suspends the use of the learned model 40. At the same time, the learning unit 65 is instructed to execute relearning (updating the learned model 40). This is because if the trained model 40 is immature due to lack of data, etc., the manual correction value will be larger than expected, so the calculation and application of the automatic correction value 32X will be stopped and the model will be improved by relearning. It is.

The display information creation unit 67 is a functional unit that is realized when the CPU 10 of the operation control device 4 executes the display information creation program 53 (FIG. 1) loaded into the main storage device 11. As shown in FIG. 3(D), the display information creation unit 67 generates the current control to be displayed on a controlled device status display screen 80, which will be described later with reference to FIG. 14, based on the operation information 24X stored in the operation information table 24. Device status information 33X representing the status of the target device 3 is created.

Further, as shown in FIG. 3(E), the display information creation unit 67 generates an automatic correction value 32X stored in an automatic correction value table 32 (described later) and operation information 24X stored in an operation information table 24 (described later). Based on this, when an abnormality in the controlled device 3 is detected, alert information 34X representing the situation of the abnormality is created.

Then, the display information creation unit 67 creates a controlled object device status display screen 80, which will be described later with reference to FIG. Screen 80 is displayed on display device 14 (FIG. 1).

On the other hand, the manufacturing information table 21 is a table for managing manufacturing information 21X (FIG. 3) regarding the product to be manufactured, and is created in advance and provided to the operation control device 4. As shown in FIG. 5, the manufacturing information table 21 includes a product column 21A, a manufacturing quantity column 21B, an equipment type column 21C, a product characteristics column 21D, a material characteristics column 21E, and a process start date and time column 21F.

The product column 21A stores an identifier (product ID) unique to the product to be manufactured, and the manufacturing quantity column 21B stores the number of products fired at one time by the controlled device 3. be done. Further, the device type column 21C stores an identifier (device type ID) representing the device type of the controlled device 3, and the product characteristics column 21D stores information related to the control of the controlled device 3, such as product size and weight. Information representing product characteristics is stored. Furthermore, the material properties column 21E stores information regarding the product materials, such as the state before processing and the raw material cost rate, and the process start date and time column 21F stores the scheduled date and time when the controlled device 3 starts manufacturing the product. Stored.

Further, the pre-process information table 22 is a table in which pre-process information 22X (FIG. 3) is stored, and is created in advance and provided to the operation control device 4. As shown in FIG. 6, this pre-process information table 22 includes a pre-process column 22A, a pre-process profile column 22B, a pre-process start date and time column 22C, a pre-process completion date and time column 22D, and a quality value column 22E. .

The pre-process column 22A stores information representing the name of the pre-process of the firing process performed in the controlled device 3, and the pre-process profile column 22B stores numerical values representing the characteristics of the work performed in the pre-process. (In the case of pressure processing, temperature, pressure, etc.) are stored. Further, the previous process start date and time column 22C stores the date and time when the previous process was started, and the previous process completion date and time column 22D stores the date and time when the previous process was completed. Furthermore, the quality value column 22E stores numerical values representing various qualities of the product, such as hardness and density, obtained during a quality inspection performed after the previous process is completed.

The physical/empirical model 23 is an algorithm generated based on a physical model consisting of a heat transfer model as described above, or an empirical model (past performance of opening a gas valve to this extent at a temperature of this extent). be.

The operation information table 24 is used to store and hold the operation information 24X (FIG. 3) of the controlled device 3 that is transmitted from the detection unit 63 (FIG. 2) of the controlled device 3 in the above-mentioned second cycle. This is a table that As shown in FIG. 7, the operation information table 24 includes an apparatus type column 24A, a time step column 24B, an acquisition date and time column 24C, a plurality of operation value columns 24D, and an environmental condition column 24E.

The device type column 24A stores the device model ID of the controlled device 3, and the time step column 24B stores the cycle at which the operation control device 4 acquires the operation information 24X (when the detection unit 63 of the controlled device 3 is in operation). This is the cycle at which the information 24X is notified, and the above-mentioned second cycle) is stored. Further, the acquisition date and time column 24C stores the date and time when the operation control device 4 acquired the operation information 24X, and the environmental condition column 24E stores the environmental conditions of the controlled device 3.

Furthermore, each operating value column 24D includes the latest value of the operating value (furnace temperature) of the controlled device 3 notified from the detection unit 63 (FIG. 2) of the controlled device 3 as operating information 24X, and the previous cycle. The current and past time-series operating values are respectively stored, such as the operating value of , the operating value of the two previous cycles, and so on.

The control history table 25 is a table used to store and hold control history information 25X (FIG. 3), which is information representing the history of control performed by the operation control device 4 or the worker on the controlled device 3. be. The latest control history information 25X is sequentially stored in the control history table 25 at the above-mentioned first cycle.

As shown in FIG. 8, this control history table 25 includes a device type column 25A, a control date and time column 25B, a control target value column 25C, an automatic correction value column 25D, a manual correction value column 25E, an operating value column 25F, and an environmental condition column. 25G. In the control history table 25, one record (row) corresponds to control history information 25X from the corresponding control target value application timing to the next control target value application timing.

The device type column 25A stores the device type ID of the controlled device 3, and the control date and time column 25B stores the date and time of the corresponding control target value application timing. Further, the control target value column 25C stores the control target value 31X (FIG. 3(A)) transmitted from the operation control device 4 to the controlled device 3 at that time. Note that the "control target value" here is a value that corresponds to the valve opening of the gas valve as described above, but if the control target value is the valve opening of the gas valve or the air valve, When a plurality of values exist, the control target value 31X also takes on a plurality of values.

The automatic correction value column 25D also contains an automatic correction value calculated by the correction information calculation unit 66 in consideration of the operating status of the controlled device 3 between the corresponding control target value application timing and the next control target value application timing. 32X (FIG. 3(C)) is stored. Note that during the period when the learning unit 65 is performing relearning (updating the learned model 40), the automatic correction value 32X is not calculated, so the automatic correction value column 25D is blank.

Further, in the manual correction value column 25E, when the worker operates the operation section 60 between the corresponding control target value application timing and the next control target value application timing, the above-mentioned information notified from the operation section 60 at that time is displayed. manual correction values are stored.

The operating value column 25F contains information from the current control target value application timing to the next control target value application timing, which is obtained as a result of transmitting the control target value 31X to the controlled device 3 at the corresponding control target value application timing. The operation value of the controlled device 3 acquired during the period is stored. Note that if there are a plurality of control target values of the control target device 3 to be controlled by the operation control device 4, all of these control target values notified from the control target device 3 are stored in the operating value column 25F.

Further, the environmental condition column 25G stores the environmental conditions of the controlled device 3 at the time when the operation control device 4 transmits the corresponding control target value 31X to the controlled device 3.

The control target value table 31 is a table used to manage the control target value 31X calculated by the control target value calculation unit 64 for each control target value application timing. As shown in FIG. 9, the control target value table 31 includes an apparatus type column 31A, a control target value application timing column 31B, and a control target value column 31C. In the control target value table 31, one record (row) corresponds to one control target value application timing at which the operation control device 4 should transmit the control target value 31X to the controlled device 3.

The device type column 31A stores the device type ID of the controlled device 3, and the control target value application timing column 31B stores the device type ID from the start of operation indicating the timing at which the control target value 31X should be transmitted to the controlled device 3. The elapsed time (seconds) is stored. Further, in the control target value column 31C, a control target value 31X to be transmitted to the controlled target device 3 at the corresponding control target value application timing is stored.

The automatic correction value table 32 is a table used to store and hold the latest automatic correction value 32X calculated by the correction information calculation unit 66 as described above, and as shown in FIG. It is configured to include an automatic correction value column 32B and an input operation information date and time column 32C.

The device type column 32A stores the device type ID of the controlled device 3, and the automatic correction value column 32B stores the automatic correction value 32X calculated last by the correction information calculation unit 66. Further, the input operating information date and time field 32C stores the acquisition date and time of the operating information 24X (FIG. 3(C)) used when calculating the automatic correction value 32X.

Further, the device status information table 33 is a table used to store and hold the above-mentioned device status information 33X (FIG. 3(D)) created by the display information creation unit 67 (FIG. 2), and is shown in FIG. It is configured to include an apparatus type field 33A, a display information date and time field 33B, an operating value field 33C, a control target value field 33D, a correction value field 33E, and an alert information field 33F. In the device status information table 33, one record (row) corresponds to one device status information 33X.

The device type column 33A stores the device type ID of the controlled device 3, and the display information date and time column 33B stores the date and time when the display information creation unit 67 created the corresponding device status information 33X. Further, the operating value column 33C stores the operating value (furnace temperature) of the controlled device 3 at the date and time when the corresponding device status information 33X is created, and the control target value column 33D stores the operating value (furnace temperature) of the controlled device 3 at the corresponding date and time. The control target value 31X applied to No. 3 is stored.

Further, the correction value column 33E stores the correction value (manual correction or automatic correction value 32X) applied to the controlled device 3 at the corresponding date and time, and the alert information column 33F stores the correction value applied to the controlled device 3 at the corresponding date and time. Information (for example, an error code) regarding an alert for an abnormality is stored. Note that if no abnormality has occurred on the corresponding date and time, the alert information column 33F will be blank.

Furthermore, the alert information table 34 is used to store and hold the above-mentioned alert information 34X (FIG. 3(E)) created when the display information creation unit 67 of the operation control device 4 detects an abnormality in the controlled device 3. As shown in FIG. 12, this table includes a device type column 34A, an alert type column 34B, an alert cause column 34C, and an alert occurrence date and time column 34D. In the alert information table 34, one record (row) corresponds to one piece of alert information 34X created when the display information creation unit 67 detects an abnormality in the controlled device 3.

The device type column 34A stores the device type ID of the controlled device 3, and the alert type column 34B stores an identifier such as an error code indicating the abnormality of the controlled device 3 detected by the display information creation unit 67 ( alert ID) is stored. Further, the alert cause column 34C stores information representing the specific cause of the abnormality, and the alert occurrence date and time column 34D stores the date and time when the abnormality occurred (more precisely, the display information creation unit 67 detected the abnormality). date and time) is stored.

(2) Flow of a series of operation control processes for the control target device in this operation control system FIG. 13 shows a flow of a series of operation control processes for the control target device 3 executed in the operation control system 1 according to the present embodiment. .

As shown in FIG. 13, in this operation control system 1, first, the manufacturing information 21X of the product stored in the manufacturing information table 21 (FIG. 5), and the Based on the process information 22X and the physical/empirical model 23, a temperature profile is created by the control target value calculation unit 64 of the operation control device 4, and the control target is determined for each control target value application timing according to the created temperature profile. The values 31X are each stored in the control target value table 31 (FIG. 9) (S1).

Each control target value 31X stored in the control target value table 31 is read out from the control target value table 31 by the control target value calculation unit 64 each time the corresponding control target value application timing arrives, and Based on the operating information 24X from the device 3, it is updated as necessary for feedforward control or feedback control, and then transmitted to the controlled device 3 (S2). At this time, the control target value calculation unit 64 sets the control target value 31X transmitted to the controlled device 3 at this time together with the transmission date and time (control date and time) as part of the control history information 25X in the control history table 25 (FIG. 8 ).

Further, the control unit 61 of the controlled target device 3 that has received the control target value 31X transmits the received control target value to the valve drive unit 62 (S3). In this way, the valve drive unit 62 drives the gas valve to open and close so that the valve opening degree is in accordance with the control target value given from the control unit 61.

Further, when the operator 70 operates the operating section 60, the control section 61 notifies the operation control device 4 of the above-mentioned manual correction value 71 notified from the operating section 60. As a result, this manual correction value 71 is registered by the correction information calculation unit 66 in the control history table 25 (FIG. 8) as part of the control history information 25X (FIG. 8).

Furthermore, the detection unit 63 of the controlled device 3 detects the latest operating value (furnace temperature) of the controlled device 3 and its acquisition date and time, the operating value of the controlled device 3 for a predetermined period in the past, and the current operating value of the controlled device 3. The information such as the environmental conditions of No. 3 is notified to the operation control device 4 in the second cycle as operation information 24X. As a result, this operating information 24X is registered in the operating information table 24 (FIG. 7) by the correction information calculation unit 66, and part of the operating information 24X (here, the latest operating values and environmental conditions) is registered in the control history table 24. (FIG. 8) is also registered.

Further, the learning unit 65 of the operation control device 4, based on the control history information 25X stored in the control history table 25, the previous process information 22X stored in the previous process information table 22, and the physical/empirical model 23, The learned model 40 is generated by learning what kind of intervention operation was performed on the controlled device 3 by the worker 70 in what state the controlled device 3 was in. Further, the learning unit 65 deploys the learned model 40 so that it can be used by the correction information calculation unit 66 when the learned model 40 reaches a certain level of maturity.

The correction information calculation unit 66 calculates the learned model 40 deployed by the learning unit 65, the previous process information 22X stored in the previous process information table 22 (FIG. 6), and the operation information 24X stored in the operation information table 24. Based on this, an automatic correction value 32X is calculated in accordance with the intervention operation on the controlled device 3 that the worker 70 would perform under the current operating conditions and operating conditions (S4).

Specifically, the correction information calculation unit 66 inputs the pre-process information 22 A correction value of 32X is calculated. Then, the correction information calculation unit 66 stores necessary information regarding the calculated automatic correction value 32X in the automatic correction value table 32 (FIG. 10) and the control history table 25.

Further, the correction information calculation unit 66 refers to the manual correction value 71 included in the latest control history information 25X registered in the control history table 25, and calculates the manual It is determined whether the correction value is within an allowable range (S5).

Then, when the manual correction value 71 is within the manual correction value allowable range, the correction information calculation unit 66 transmits the automatic correction value 32X stored in the automatic correction value table 32 to the controlled device 3 (S6 ). Further, the correction information calculation unit 66 converts the automatic correction value 32X sent to the controlled device 3 at this time into the automatic correction value column 25D ( Figure 8).

On the other hand, if the manual correction value 71 is outside the allowable manual correction value range, the correction information calculation unit 66 temporarily stops using the learned model 40 and updates the learned model 40. An instruction is given to the learning unit 65 to start or restart (re-learning) (S7).

On the other hand, the display information creation unit 67 generates device status information 33X based on the operation information 24X stored in the operation information table 24, and stores the generated device status information 33X in the device status information table 33 (FIG. 11). (S8). The display information creation unit 67 also creates a controlled device status display screen 80, which will be described later with reference to FIG. 14, based on the device status information 33X stored in the device status information table 33, and displays the created controlled device status display screen 80. It is displayed on the device 14 (S10).

Further, the display information creation unit 67 constantly determines whether there is an abnormality in the controlled device 3 based on the automatic correction value 32X calculated by the correction information calculation unit 66. This is because when the tendency of the controlled device 3 changes due to aging etc., the dispersion of the automatic correction value 32X calculated by the correction information calculation unit 66 becomes large, the automatic correction value 32X itself becomes large, or the automatic This is because the fluctuation of the correction value 32X becomes large, so by monitoring the automatic correction value 32X, it is possible to determine whether or not an abnormality has occurred in the controlled device 3.

The display information creation unit 67 determines whether any of the variations in the automatic correction values 32X, the size of the automatic correction values 32X itself, or the fluctuations in the automatic correction values 32X If the size of 32X itself and the automatic correction value 32X exceed the predetermined ranges (hereinafter collectively referred to as the automatic correction value tolerance range), physical /Analyze the factors that caused the automatic correction value 32X to be outside the automatic correction value tolerance range (factors that caused the abnormality) based on the degree of deviation from the empirical model 23, generate alert information 34X based on the analysis results, and generate alert information. It is stored in the table 34 (FIG. 12) (S9). The display information creation unit 67 also displays the abnormal situation of the controlled device 3 on the controlled device status display screen 80 based on the alert information 34X stored in the alert information table 34 (S10).

(3) Configuration of Controlled Device Status Display Screen FIG. 14 shows a configuration example of the controlled device status display screen 80 created by the display information creation section 67 and displayed on the display device 14. As shown in FIG. 14, it is configured to include an operating status display area 81 provided on the left side of the screen and an abnormal status display area 82 provided on the right side of the screen.

The operating status display area 81 is an area for displaying the operating status of the controlled device 3. While the controlled device 3 is in operation, the operating status display area 81 is an area for displaying the operating status of the controlled device 3. Changes in the operating value (furnace temperature) in the controlled device 3 are constantly displayed in a graph or the like. Note that FIG. 14 exemplifies a case where not only the temperature inside the furnace but also the water temperature of the "xx water tank" is to be monitored and controlled.

Further, the abnormal situation display area 82 includes a summary display area 82A and a probable cause display area 82B. In the summary display area, when the display information creation unit 67 detects an abnormality in the controlled device 3, the display information creation unit 67 displays a summary of the abnormality based on the alert information 34X stored in the alert information table 34. be done.

Further, in the estimated cause display area 82B, several candidates (hereinafter referred to as cause candidates) estimated by the display information creation unit 67 as causes of the abnormality are displayed in a list in a cause candidate list 83. In this cause candidate list 83, for each cause candidate, a character string 83A representing the specific content and a level meter 83B representing the probability that the cause candidate is the cause of the abnormality are displayed.

The number of cause candidates to be displayed in the cause candidate list 83 (hereinafter referred to as the number of displayed cause candidates) can be determined by clicking the pull-down button 84 displayed above the cause candidate list 83. This can be specified by selecting a number that matches the desired number of display cause candidates from among the numbers displayed on a menu (not shown). Then, the number of display cause candidates selected at this time is displayed in a text box 85 provided on the left side of the pull-down button 84.

(4) Various processes executed by each functional unit of the operation control device Next, specific processing contents of various processes executed in the operation control device 4 to control the operation of the controlled device 3 will be described. In the following, the processing main body of each process will be explained as a functional section ("... section"), but in reality, the CPU 10 (FIG. 1) of the operation control device 4 executes the processing based on the program corresponding to the functional section. Needless to say, the process is executed.

(4-1) Operation control processing FIG. 15 shows the operation control processing executed by the control target value calculation unit 64 and the correction information calculation unit 66 of the operation control device 4 in cooperation to control the operation of the controlled device 3. Show the flow. The control target value calculation unit 64 and the correction information calculation unit 66 control the operation of the controlled device 3 according to the processing procedure shown in FIG.

In practice, in this operation control process, after a new material to be fired is input into the controlled device 3, an operation input by the worker 70 to start the operational control of the controlled device 3 is sent to the operation control device 4. (FIG. 13).

First, the control target value calculation unit 64 uses the manufacturing information 21X stored in the manufacturing information table 21 (FIG. 5), the pre-process information 22X stored in the pre-process information table 22 (FIG. 6), and the physical/ Using the empirical model 23, a temperature profile of the furnace temperature in the controlled device 3 is drawn up, and the control target value 31X for each control target value application timing based on the designed temperature profile is set in the control target value table 31 (FIG. 9). ) (S20).

Subsequently, the correction information calculation unit 66 determines whether the learned model 40 has been deployed (S21). If the correction information calculation unit 66 obtains a negative result in this judgment, the process proceeds to step S23, and if a positive result is obtained, the correction information calculation unit 66 calculates the previous process information 22X stored in the previous process information table 22 and the operation information table 24 (FIG. 7). The automatic correction value 32X (FIG. 3(C)) is calculated by inputting the operation information 24X stored in the learned model 40 into the learned model 40. Then, the correction information calculation unit 66 stores the calculated automatic correction value 32X in the control history table 25 and the automatic correction value table 32, and transmits it to the controlled device 3 (S22). As a result, in the controlled device 3, the control target value 31X is corrected based on this automatic correction value 32X, and the furnace temperature (gas valve opening) in the controlled device 3 is adjusted based on the corrected control target value 31X. degree) control is performed.

Next, the correction information calculation unit 66 determines whether the latest control history information 25X stored in the control history table 25 (FIG. 8) includes the manual correction value 71 (FIG. 13) (S23).

Obtaining a negative result in this judgment means that the operator 70 (see FIG. ) means that no intervention operation has been performed. Thus, at this time, the correction information calculation section 66 proceeds to step S28.

On the other hand, obtaining a positive result in the determination in step S23 means that the correction information calculation unit 66 last sends the control target value to the controlled device 3 until it acquires the latest control history information 25X. This means that the operator 70 performed an intervention operation. Thus, at this time, the correction information calculation unit 66 determines that the manual correction value 71 (FIG. 13) output from the operation unit 60 (FIG. 2) of the controlled device 3 as a result of the intervention operation is within the manual correction value tolerance range. It is determined whether or not (S24).

Obtaining a negative result in step S24 means that the control of the furnace temperature based on the control target value 31X transmitted from the operation control device 4 to the controlled device 3 was insufficient, and the worker 70 (see FIG. ) means that a large intervention operation was performed. Thus, at this time, the correction information calculation unit 66 stops calculation of the automatic correction value 32X using the learned model 40 and correction of the control target value 31X using the automatic correction value 32X (S25). As a result, the control target value 31X stored in the control target value table 31 will be sent to the controlled device 3 without being corrected, and the controlled target device 3 will be notified from the operation unit 60. Operation control of the controlled device 3 is performed taking into consideration only the manual target value.

Also, at this time, the correction information calculation section 66 instructs the learning section 65 to re-learn (hereinafter, this instruction will be referred to as a re-learning instruction) (S26). As a result, the learning unit 65 updates (re-learning) the trained model 40 according to the processing procedure described later with reference to FIG. Ru.

On the other hand, obtaining a positive result in the determination in step S24 means that the control of the furnace temperature based on the control target value 31X transmitted from the operation control device 4 to the controlled device 3 was sufficient. (FIG. 13) means that no major intervention operation was performed. Thus, at this time, the correction information calculation unit 66 starts or resumes calculation of the automatic correction value 32X using the learned model 40 and use of the calculated automatic correction value 32X (S27).

Subsequently, the correction information calculation section 66 acquires the operation information 24X (FIG. 3) transmitted from the detection section 63 (FIG. 2) of the controlled device 3, and stores the acquired operation information 24X in the operation information table 24 (FIG. 7) (S28), and it is determined whether the current firing process in the controlled device 3 has been completed (S29).

When the correction information calculation unit 66 obtains a negative result in this determination, it returns to step S21, and thereafter repeats the processing from step S21 to step S29 until it obtains a positive result in step S28. Then, when the current firing process in the controlled device 3 is completed, the correction information calculation unit 66 ends this operation control process.

(4-2) Relearning Process FIG. 16 shows the relearning process executed by the learning unit 65 to which the relearning instruction is given from the correction information calculation unit 66 in step S26 of the operation control process described above with reference to FIG. In accordance with the processing procedure shown in FIG. 16, the learning section 65 determines when the temperature inside the furnace of the controlled device 3 and its surrounding environment are such that the operator 70 (FIG. 13) can control the controlled object via the operating section 60. Re-learning what kind of intervention operation was performed on the device 3.

In practice, when the learning unit 65 receives a relearning instruction from the correction information calculation unit 66, it starts the learning process shown in FIG. Among the information 25X, data of the control history information 25X, which has not yet been used for learning the learned model 40, is input to carry out such relearning to update the learned model 40 (S30). Further, the learning unit 65 then deploys the updated learned model 40 so that the correction information calculation unit 66 can use it (S31), and then ends this relearning process.

(4-3) Screen creation and display processing FIG. 17 shows the processing procedure of the screen creation and display processing executed by the display information creation section 67. The display information creation unit 67 creates the controlled device status display screen 80 described above with reference to FIG. 14 according to the processing procedure shown in FIG. 17, and displays the created controlled device status display screen 80 on the display device 14 (FIG. 1). do.

In fact, when the operation control device 4 is started, the display information creation unit 67 starts the screen creation and display process shown in FIG. Either the variation in the automatic correction value 32X or the variation in the automatic correction value 32X is related to the size of the automatic correction value 32X itself, the variation in the automatic correction value 32X, or the variation in the automatic correction value 32X. It is determined whether each of the values falls within a predetermined range (hereinafter referred to as an automatic correction value tolerance range) (S40).

If the display information creation unit 67 obtains a positive result in this judgment, the process proceeds to step S43, and displays the operating status of the controlled device 3 to be displayed in the operating status display area 81 (FIG. 14) on the controlled device status display screen 80. The generated device status information 33X (FIG. 11) is generated, and the generated device status information 33X is stored in the device status information table 33 (FIG. 11) (S43).

In addition, the display information creation unit 67 generates a controlled device status display screen 80 on which information (a graph in FIG. 14) representing the operating status of the controlled device 3 is posted based on the device status information 33X stored in the device status information table 33. Create. Further, the display information creation unit 67 displays the created controlled device status display screen 80 on the display device 14 (S44), and then returns to step S40 to repeat the processes from step S40 to step S44.

On the other hand, when the display information creation unit 67 obtains a negative result in the determination in step S40, it determines that an abnormality has occurred in the controlled device 3, and analyzes the cause of the abnormality (S41). As a method, there is a method that utilizes the deviation from the physical model regarding the size of the automatic correction value 32X itself that is outside the corresponding automatic correction value tolerance range, the dispersion of the automatic correction value 32X, and/or the fluctuation of the automatic correction value 32X. be.

Subsequently, the display information creation unit 67 generates alert information 34X (FIG. 12) based on the analysis result of step S41, and stores the generated alert information 34X in the alert information table 34 (FIG. 12) (S42). Further, the display information creation unit 67 generates the device status information 33X in the same manner as described above, and stores the generated device status information 33X in the device status information table 33 (S43).

Then, the display information creation unit 67 creates the controlled device status display screen 80 based on the device status information 33X stored in the device status information table 33 and the alert information 34X stored in the alert information table 34, and also creates the The controlled device status display screen 80 is displayed on the display device 14 (S44). Further, the display information creation unit 67 returns to step S40, and repeats the processes from step S40 to step S44.

(5) Effects of the present embodiment As described above, in the operation control system 1 of the present embodiment, the operation control device 4 determines when the operating value and environmental conditions of the controlled device 3 are 70 (FIG. 13) learns what kind of intervention operation has been performed on the controlled device 3 via the operation unit 60, and uses the learned model 40 obtained by learning to change the current controlled device The automatic correction value 32X is calculated by estimating what kind of intervention operation the worker 70 (FIG. 13) will perform on the controlled device 3 when the operating value and environmental conditions are 3. 32X to control the controlled device 3.

Therefore, according to the present driving control system 1, it is possible to perform control that coordinates automatic control and human intervention, and it is possible to perform control that is closer to the sensibilities of an expert.

(6) Other Embodiments In the above-described embodiments, a case has been described in which the present invention is applied to controlling the furnace temperature in the firing process of carbon electrodes, but the present invention is not limited to this. First, it can be widely applied to control various control objects, such as controlling the screw rotation speed and/or cylinder temperature in the extrusion molding process of resin pellets, and controlling the temperature and/or humidity inside a plant factory. .

Furthermore, in the above-described embodiment, the first cycle (cycle of control target value application timing) in which the operation control device 4 transmits the control target value 31X to the control target device 3, and As described above, the second cycle for transmitting the automatic correction value 32X (the cycle for the detection unit 63 of the controlled device 3 to transmit information such as operating values and environmental conditions to the operation control device 4) is different. However, the present invention is not limited to this, and the first and second cycles may be the same cycle and the same timing. In this case, the operation control device 4 may correct the control target value 31X using the calculated automatic correction value 32X, and then transmit the corrected control target value 31X to the controlled device 3.

Furthermore, in the above-described embodiment, a case has been described in which the operation control device 4 is constituted by one computer device, but the present invention is not limited to this, and the present invention is applicable to distributed computing constituted by a plurality of computer devices. It may be configured by the system.

Further, in the above-described embodiment, a case has been described in which the controlled device status display screen 80 is configured as shown in FIG. can do.

Furthermore, in the embodiment described above, the control target device 3 is adjusted based on the magnitude of the automatic correction value 32X itself calculated by the correction information calculation unit 66, the dispersion of the automatic correction value 32X, and the fluctuation of the automatic correction value 32X. Although the case has been described in which the presence or absence of an abnormality is determined and the cause of the abnormality is analyzed, the present invention is not limited to this. Alternatively, the cause of the abnormality may be analyzed.

The present invention can be widely applied to various control devices that control a controlled object to a predetermined state.

1... Operation control system, 3... Controlled device, 4... Operation control device, 10... CPU, 14... Display device, 21... Manufacturing information table, 21X... Manufacturing information, 22... Pre-process Information table, 22X...previous process information, 23...physical/empirical model, 24...operation information table, 24X...operation information, 25...control history table, 25X...control history information, 31...control target Value table, 31X...Control target value, 32...Auto correction value table, 32X...Auto correction value, 33...Device status information table, 33X...Device status information, 34...Alert information table, 34X... Alert information, 40... Learned model, 64... Control target value calculation section, 65... Learning section, 66... Correction information calculation section, 67... Display information creation section, 70... Worker, 71... Manual correction value, 80...Controlled device status display screen.

Claims (14)

In a control device that controls a controlled object to a predetermined state,
a control target calculation unit that controls the controlled object by calculating a control target for controlling the controlled object to the predetermined state and transmitting the calculated control target to the controlled object;
A first correction amount by a manual intervention operation on the control object and a control result of the control object when the intervention operation is performed, and the obtained first correction amount and the control result of the control object a learning unit that learns the first correction amount for the control result of the controlled object based on the control result;
Acquire the current control result of the controlled object, and calculate a second correction amount for the control target using the acquired control result of the controlled object and the learned model obtained by learning by the learning unit. and a correction information calculation section for
The correction information calculation unit includes:
A control device comprising: correcting the control target using the calculated second correction amount; or transmitting the calculated second correction amount to the control target. The learning department is
acquiring the current and past time-series control results of the controlled object, and learning the first correction amount for the control result of the controlled object based on the acquired time-series control results of the controlled object; The control device according to claim 1, characterized in that: The learning department is
learning the first correction amount for the control result of the controlled object when a manual intervention operation is performed based on the amount of change in the control result of the controlled object;
The correction information calculation unit includes:
A current control result of the controlled object is acquired, and the second correction amount for the control target is calculated using the learned model based on the amount of change in the acquired control result of the controlled object. The control device according to claim 1 or claim 2. The correction information calculation unit includes:
A claim characterized in that the calculation of the second correction amount is stopped when the first correction amount exceeds a predetermined first range set in advance for the first correction amount. The control device according to item 1. The learning department is
The control device according to claim 4, wherein when the first correction amount exceeds the first range, the first correction amount for the control result of the controlled object is re-learned. It is characterized by further comprising a display unit that determines the presence or absence of an abnormality in the control device based on the control result of the controlled object, and when an abnormality is detected, analyzes the cause of the abnormality and displays the analysis result. The control device according to claim 1. The display section is
Any of the variation in the second correction amount, the size of the second correction amount itself, and/or the variation in the second correction amount is the variation in the second correction amount, the second correction amount itself, If the magnitude and/or fluctuation of the second correction amount exceeds a predetermined second range set in advance, it is assumed that an abnormality has occurred in the controlled object, and the second correction amount is determined to be abnormal. 7. The control device according to claim 6, wherein the control device analyzes the cause of the deviation outside the second range and displays the analysis result. A control method executed by a control device that controls a controlled object to a predetermined state, the method comprising:
a first step of controlling the controlled object by calculating a control target for controlling the controlled object to the predetermined state and transmitting the calculated control target to the controlled object;
A first correction amount by a manual intervention operation on the control object and a control result of the control object when the intervention operation is performed, and the obtained first correction amount and the control result of the control object a second step of learning the first correction amount for the control result of the controlled object based on the control result;
Acquire the current control result of the controlled object, and calculate a second correction amount for the control target using the acquired control result of the controlled object and the learned model obtained by learning by the learning unit. and a third step to
In the third step, the control device:
A control method comprising: correcting the control target using the calculated second correction amount; or transmitting the calculated second correction amount to the control target. In the second step, the control device:
acquiring the current and past time-series control results of the controlled object, and learning the first correction amount for the control result of the controlled object based on the acquired time-series control results of the controlled object; The control method according to claim 8, characterized in that: In the second step, the control device:
learning the first correction amount based on the amount of change in the control result of the controlled object so that the control result of the controlled object when a manual intervention operation is performed is within a predetermined range;
In the third step, the control device:
A current control result of the controlled object is acquired, and the second correction amount for the control target is calculated using the learned model based on the amount of change in the acquired control result of the controlled object. The control method according to claim 8 or claim 9. In the third step, the control device:
A claim characterized in that the calculation of the second correction amount is stopped when the first correction amount exceeds a predetermined first range set in advance for the first correction amount. The control method according to item 8. In the second step, the control device:
The control method according to claim 11, characterized in that, when the first correction amount exceeds the first range, the first correction amount for the control result of the controlled object is re-learned. In the third step, the control device:
According to claim 8, the control device determines whether or not there is an abnormality in the control device based on the control result of the controlled object, and when an abnormality is detected, analyzes the cause of the abnormality and displays the analysis result. Control method described. In the third step, the control device:
Any of the variation in the second correction amount, the size of the second correction amount itself, and/or the variation in the second correction amount is the variation in the second correction amount, the second correction amount itself, If the magnitude and/or fluctuation of the second correction amount exceeds a predetermined second range set in advance, it is assumed that an abnormality has occurred in the controlled object, and the second correction amount is determined to be abnormal. 14. The control method according to claim 13, further comprising analyzing the cause of the deviation outside the second range and displaying the analysis result.

Images