Control device, control method, and recording medium having control program recorded thereon

Abstract

A control device, a control method, and a recording medium having a control program recorded thereon are provided. The control device comprises: a control unit that controls the control object using a control model that has been machine-learned in such a manner that an operation amount of the control object is output in accordance with a state of a device provided with the control object; a prediction unit that predicts a future state of the device; an adjustment unit that adjusts the operation amount based on a prediction result; and an output unit that outputs the adjusted operation amount to the control object.

Description

Technical Field

[0001] The present invention relates to a control device, a control method, and a recording medium recording a control program. Background Art

[0002] Patent document 1 states that "machine learning is performed on the correction amount of the robot's teaching position in response to the interference applied to the motors driving the joints of the robot, and based on the results of the machine learning, the robot is controlled while correcting the teaching position in a manner that suppresses the interference as it moves toward the teaching position."

[0003] Patent Document 1: Japanese Patent Application Laid-Open No. 2018-202564 Summary of the Invention

[0004] (Item 1)

[0005] In a first aspect of the present invention, a control device is provided. The control device may include a control unit that controls a controlled object using a control model obtained through machine learning in a manner that outputs an operation variable of the controlled object in accordance with the state of the device in which the controlled object is installed. The control device may include a prediction unit that predicts the future state of the device. The control device may include an adjustment unit that adjusts the operation variable based on the prediction result. The control device may include an output unit that outputs the adjusted operation variable to the controlled object.

[0006] (Item 2)

[0007] The prediction unit can predict the future health of the equipment, and the adjustment unit can adjust the operation amount according to the health.

[0008] (Item 3)

[0009] Soundness may include the remaining wall thickness of the piping of the equipment.

[0010] (Item 4)

[0011] The forecasting unit can predict the future maintenance cost of the equipment, and the adjustment unit can adjust the operation volume according to the maintenance cost.

[0012] (Item 5)

[0013] Maintenance costs may include costs associated with the addition of inhibitors to inhibit degradation of the equipment.

[0014] (Item 6)

[0015] The control device may further include a learning unit that generates a control model through machine learning.

[0016] (Item 7)

[0017] The learning unit may generate the control model by performing reinforcement learning in response to input of state data indicating the state of the device so as to output a more recommended operation amount for an operation amount having a higher reward value determined by a predetermined reward function.

[0018] (Item 8)

[0019] The learning unit may also learn the constraints of machine learning based on the history of adjusting the operation amount.

[0020] (Item 9)

[0021] The constraint condition may include at least one of an upper limit value and a lower limit value of the operation amount.

[0022] (Item 10)

[0023] The control device may further include a plan acquisition unit that acquires a production plan of the equipment. The adjustment unit may adjust the operation amount based on the prediction result and the production plan.

[0024] (Item 11)

[0025] In a second aspect of the present invention, a control method is provided. The control method may include the step of controlling a controlled object using a control model obtained through machine learning in a manner that outputs an operation variable of the controlled object in accordance with the state of the device in which the controlled object is installed. The control method may include the step of predicting the future state of the device. The control method may include the step of adjusting the operation variable based on the prediction result. The control method may include the step of outputting the adjusted operation variable to the controlled object.

[0026] (Item 12)

[0027] In a third aspect of the present invention, a recording medium is provided, the recording medium having a control program recorded thereon. The control program can be executed by a computer. The control program can cause the computer to function as a control unit that controls the control object using a control model obtained through machine learning in a manner that outputs the manipulated variable of the control object in accordance with the state of the device provided with the control object. The control program can cause the computer to function as a prediction unit that predicts the future state of the device. The control program can cause the computer to function as an adjustment unit that adjusts the manipulated variable based on the prediction result. The control program can cause the computer to function as an output unit that outputs the adjusted manipulated variable to the control object.

[0028] The above summary of the invention does not list all the features of the present invention. In addition, sub-components of the above-mentioned feature groups can also constitute the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 An example of a block diagram of the device 10 provided with the control target 20 is shown together with the control apparatus 100 according to the present embodiment.

[0030] Figure 2 An example of a flow in which the control device 100 according to the present embodiment generates the control model 135 through machine learning is shown.

[0031] Figure 3 An example of a flow of adjusting the operation amount by the control device 100 according to the present embodiment is shown.

[0032] Figure 4 An example of a block diagram of a device 10 provided with a control target 20 is shown together with a control apparatus 100 according to a first modification of the present embodiment.

[0033] Figure 5 An example of a block diagram of a device 10 provided with a control target 20 is shown together with a control apparatus 100 according to a second modification of the present embodiment.

[0034] Figure 6 An example of a flow of adjusting the control amount by the control device 100 according to the second modified example of the present embodiment is shown.

[0035] Figure 7 An example of a computer 9900 is shown that can embody all or part of the various aspects of the present invention. DETAILED DESCRIPTION

[0036] The present invention will be described below by way of embodiments of the invention, but the following embodiments do not limit the invention as defined in the claims. Furthermore, all combinations of features described in the embodiments are not necessarily essential for the solution to the problem.

[0037] Figure 1 An example of a block diagram of a device 10 equipped with a controlled object 20 is shown together with a control device 100 according to this embodiment. The control device 100 according to this embodiment predicts the future state of the device 10 equipped with the controlled object 20 when controlling the controlled object 20 using a learning model generated through machine learning (artificial intelligence, also known as AI control). Furthermore, the control device 100 according to this embodiment adjusts the manipulated variable output by the learning model based on the prediction results.

[0038] Equipment 10 is a facility, device, or the like equipped with a controlled object 20. For example, equipment 10 may be a factory or a complex device composed of multiple instruments. Examples of factories include, in addition to chemical and biological industrial plants, plants that manage and control the wellheads of gas and oil fields and their surrounding areas; plants that manage and control hydropower, thermal power, and nuclear power generation; plants that manage and control environmental resource generation such as solar and wind power; and plants that manage and control water supply and drainage systems, dams, and the like.

[0039] A control target 20 is provided for the device 10. In this figure, a case where only one control target 20 is provided for the device 10 is shown as an example, but the present invention is not limited to this. A plurality of control targets 20 may be provided for the device 10.

[0040] Furthermore, the device 10 may be provided with one or more sensors (not shown) that measure various states (physical quantities) inside and outside the device 10. Such sensors measure, for example, operating data, consumption data, and external environment data.

[0041] Here, the operating data represents the operating state resulting from controlling the controlled object 20. For example, the operating data may represent a measured value PV (Process Variable) measured for the controlled object 20, and as an example, may represent the output (controlled variable) of the controlled object 20, or may represent various values ​​that change depending on the output of the controlled object 20.

[0042] The consumption data indicates the consumption of at least one of energy and raw materials by the equipment 10. For example, the consumption data may indicate the consumption of electricity or fuel (for example, LPG: Liquefied Petroleum Gas) as energy consumption.

[0043] External environment data represents physical quantities that can interfere with the control of control object 20. For example, external environment data may represent the temperature, humidity, sunshine, wind direction, wind volume, precipitation, and other physical quantities of the atmosphere outside device 10 that vary with the control of other devices installed in device 10.

[0044] Controlled object 20 is an instrument or device that is controlled. For example, controlled object 20 may be an actuator such as a valve, pump, heater, fan, motor, or switch that controls at least one physical quantity of the process in equipment 10, such as pressure, temperature, pH, velocity, or flow rate. It receives a manipulated variable (MV) as input and outputs a controlled variable.

[0045] The control device 100 according to this embodiment predicts the future state of the device 10 provided with the control target 20 when performing AI control on the control target 20. The control device 100 according to this embodiment adjusts the operation amount output by the learning model based on the prediction result.

[0046] The control device 100 can be a computer such as a PC (personal computer), a tablet computer, a smart phone, a workstation, a server computer, or a general-purpose computer, or a computer system formed by connecting multiple computers. In addition, this computer system is also a computer in a broad sense. In addition, the control device 100 can be installed in a computer using one or more executable virtual computer environments. Alternatively, the control device 100 can be a dedicated computer designed for AI control, or it can be dedicated hardware implemented using dedicated circuits. In addition, when the control device 100 can be connected to the Internet, the control device 100 can be implemented through cloud computing.

[0047] The control device 100 includes a state data acquisition unit 110, an operation variable data acquisition unit 120, a learning unit 130, a control model 135, a control unit 140, a prediction unit 150, an adjustment unit 160, and an output unit 170. Furthermore, these modules may be functionally separate modules and may not correspond to the actual device structure. That is, although they are shown as a single module in this figure, they do not need to be composed of a single device. Furthermore, although they are shown as separate modules in this figure, they do not need to be composed of separate devices.

[0048] The status data acquisition unit 110 acquires status data representing the status of the device 10 in which the control target 20 is installed. For example, the status data acquisition unit 110 acquires operating data, consumption data, and external environment data measured by sensors installed in the device 10 via a network. However, this is not limiting. The status data acquisition unit 110 may acquire this status data from an operator or from various memory devices. The status data acquisition unit 110 supplies the acquired status data to the learning unit 130, the control model 135, and the prediction unit 150.

[0049] The manipulated variable data acquisition unit 120 acquires manipulated variable data representing the manipulated variable of the controlled object 20. For example, the manipulated variable data acquisition unit 120 acquires, from the control unit 140, data representing the manipulated variable MV(AI) output by the control model 135 when AI control is performed on the controlled object 20 as the manipulated variable data. However, this is not a limitation. The manipulated variable data acquisition unit 120 may acquire such manipulated variable data from the operator or from various memory devices. The manipulated variable data acquisition unit 120 supplies the acquired manipulated variable data to the learning unit 130 and the adjustment unit 160.

[0050] Furthermore, this figure illustrates, as an example, a case where the manipulated variable data acquisition unit 120 acquires data representing the manipulated variable MV(AI) output by the control model 135 as manipulated variable data. However, this is not limiting. During the learning phase, when the control device 100 performs machine learning using data from the control object 20 controlled by another controller (not shown) as learning data, the manipulated variable data acquisition unit 120 may acquire data representing the manipulated variable applied to the controlled object 20 from the other controller as manipulated variable data. As an example, when the controlled object 20 is capable of switching between feedback control based on the manipulated variable MV (FB) applied by the other controller and AI control based on the manipulated variable MV(AI) applied by the control model 135, the manipulated variable data acquisition unit 120 may acquire data representing the manipulated variable MV(FB) applied to the controlled object 20 from the other controller during the learning phase as manipulated variable data. Furthermore, this FB control may be, for example, control utilizing at least one of proportional control (P control), integral control (I control), or differential control (D control), and may also be PID control. In addition, such another controller may be integrally configured as a part of the control device 100 according to the present embodiment, or may be configured as a separate structure independent of the control device 100 .

[0051] The learning unit 130 uses state data and manipulated variable data through machine learning to generate a control model 135 that outputs manipulated variables corresponding to the state of the device 10. For example, the learning unit 130 performs reinforcement learning using the state data supplied by the state data acquisition unit 110 and the data representing the manipulated variable MV (AI) supplied by the manipulated variable data acquisition unit 120 as learning data, thereby generating a control model 135 that outputs the manipulated variable MV (AI) corresponding to the state of the device 10. Specifically, the learning unit 130 generates the control model 135 by performing reinforcement learning based on the input state data so that the manipulated variable with a higher reward value specified by a predetermined reward function is output as a more recommended manipulated variable. This will be described in detail later.

[0052] The control model 135 is a learning model generated by the learning unit 130 through machine learning, and outputs the operation amount MV (AI) corresponding to the state of the device 10. For example, the control model 135 inputs the state data supplied from the state data acquisition unit 110, and outputs the recommended operation amount MV (AI) that should be given to the control object 20 based on the state of the device 10. In addition, in this figure, as an example, the control model 135 is shown as being built into the control device 100, but it is not limited to this. The control model 135 can be stored in a device different from the control device 100 (for example, a cloud server). Similarly, the learning unit 130 can also be equipped with a device different from the control device 100.

[0053] The control unit 140 controls the controlled object 20 using the manipulated variable MV (AI) output by the control model 135. Specifically, the control unit 140 controls the controlled object 20 using the control model 135, which is obtained through machine learning in a manner that outputs the manipulated variable of the controlled object 20 according to the state of the device 10 in which the controlled object 20 is installed. Furthermore, the control unit 140 supplies the manipulated variable MV (AI) output by the control model 135 to the manipulated variable data acquisition unit 120.

[0054] The prediction unit 150 predicts the future state of the device 10. This prediction is based on the state data supplied by the state data acquisition unit 110. The term "prediction" here encompasses not only the prediction unit 150 itself predicting the future state of the device 10, but also the prediction unit 150 predicting the future state of the device 10 for other devices and acquiring the predicted future state of the device 10 from other devices. The prediction unit 150 supplies the prediction results of the future state of the device 10 to the adjustment unit 160.

[0055] During the learning phase, adjustment unit 160 directly supplies the manipulated variable MV(AI) supplied from manipulated variable data acquisition unit 120 to output unit 170. Meanwhile, during the operational phase, adjustment unit 160 adjusts manipulated variable MV(AI) based on the prediction results supplied as needed from prediction unit 150. Adjustment unit 160 then supplies the adjusted manipulated variable MV(AI)_adj to output unit 170.

[0056] The output unit 170 outputs the manipulated variable MV supplied from the adjustment unit 160 to the controlled object 20. Specifically, during the learning phase, the output unit 170 outputs the manipulated variable MV(AI) output by the control model 135 to the controlled object 20. Meanwhile, during the operational phase, the output unit 170 outputs the manipulated variable MV(AI) output by the control model 135 or the manipulated variable MV(AI)_adj adjusted by the adjustment unit 160 to the controlled object 20. This will be described in detail using a flowchart.

[0057] Figure 2 The following is an example of a flow in which the control device 100 according to this embodiment generates the control model 135 through machine learning. In the learning phase, the control device 100 generates the control model 135 that outputs the manipulated variable according to the state of the device 10 through machine learning using state data and manipulated variable data.

[0058] In step 210, the control device 100 acquires status data. For example, the status data acquisition unit 110 acquires status data indicating the status of the device 10 in which the control target 20 is installed. As an example, the status data acquisition unit 110 acquires operating data, consumption data, and external environment data measured by sensors installed in the device 10 via a network as status data. The status data acquisition unit 110 supplies the acquired status data to the learning unit 130 and the control model 135.

[0059] In step 220, the control device 100 acquires manipulated variable data. For example, the manipulated variable data acquisition unit 120 acquires manipulated variable data representing the manipulated variable of the controlled object 20. As an example, the manipulated variable data acquisition unit 120 acquires, from the control unit 140, data representing the manipulated variable MV(AI) output by the control model 135 when AI control is performed on the controlled object 20 as the manipulated variable data. The manipulated variable data acquisition unit 120 supplies the acquired manipulated variable data to the learning unit 130 and the adjustment unit 160. Accordingly, the adjustment unit 160 directly supplies the manipulated variable MV(AI) supplied from the manipulated variable data acquisition unit 120 to the output unit 170. Furthermore, the output unit 170 outputs the manipulated variable supplied from the adjustment unit 160 to the controlled object 20. That is, during the learning phase, the output unit 170 outputs the manipulated variable MV(AI) output by the control model 135 to the controlled object 20. While this figure illustrates an example in which the control device 100 acquires manipulated variable data after acquiring state data, this is not limiting. The control device 100 may acquire the state data after acquiring the operation amount data, or may acquire the state data and the operation amount data at the same time.

[0060] In step 230, the control device 100 generates a control model 135. For example, the learning unit 130 uses the state data and the manipulated variable data to generate the control model 135 by machine learning, thereby outputting the manipulated variable corresponding to the state of the device 10. As an example, the learning unit 130 performs reinforcement learning using the state data acquired in step 210 and the manipulated variable data representing the manipulated variable MV (AI) acquired in step 220 as learning data, thereby generating the control model 135 that outputs the manipulated variable MV (AI) corresponding to the state of the device 10.

[0061] Typically, if an agent observes the state of the environment and chooses an action, the environment changes based on that action. In reinforcement learning, by giving a certain reward along with changes in the environment, the agent learns to choose a better action (willingness decision). In contrast to learning with a teacher, which can obtain completely correct answers, reinforcement learning gives rewards as discontinuous values ​​based on changes in part of the environment. Therefore, the agent learns to choose the action that will maximize the total reward in the future. In this way, in reinforcement learning, the agent learns appropriate actions based on the interaction of learning actions and taking actions on the environment, that is, learning actions to maximize the reward received in the future.

[0062] In this embodiment, the reward for reinforcement learning can be an indicator used to evaluate the operation of the device 10, or a value determined by a predetermined reward function. Here, a function refers to a mapping with a rule that establishes a one-to-one correspondence between elements of another set and elements of a particular set. For example, it can be a mathematical formula or a table.

[0063] The reward function outputs a value (reward value) that evaluates the state of the device 10 represented by the state data based on the input of the state data. As described above, for example, the state data includes the measured value PV measured for the control object 20. Therefore, the reward function can be defined as a function in which the closer the measured value PV is to the target value SV (Setting Variable), the higher the reward value. Here, the evaluation function can be defined as a function with the absolute value of the difference between the measured value PV and the target value SV as a variable. That is, as an example, when the control object 20 is a valve, the evaluation function can be a function with the absolute value of the difference between the valve opening actually measured by the sensor, that is, the measured value PV, and the valve opening set as the target, that is, the target value SV, as a variable. Furthermore, the reward function can be a function with the value of the evaluation function obtained by this evaluation function as a variable.

[0064] Furthermore, as described above, in addition to the measured value PV, the state data also includes, for example, various values ​​that change depending on the output of the controlled object 20, consumption data, external environmental data, and the like. Therefore, the reward function can be a function that increases or decreases the reward value based on these various values, consumption data, and external environmental data. As an example, if constraints are imposed on these various values ​​or consumption data that must be adhered to, the reward function can be a function that minimizes the reward value if, with reference to the external environmental data, these various values ​​or consumption data do not meet the constraints. Furthermore, if targets are imposed on these various values ​​or consumption data that must be achieved, the reward function can be a function that increases the reward value as these various values ​​or consumption data approach the targets, and decreases the reward value as they move further away from the targets, with reference to the external environmental data.

[0065] The learning unit 130 obtains a reward value for each learning data item based on this reward function. Furthermore, the learning unit 130 performs reinforcement learning using each set of learning data and reward values. In this case, the learning unit 130 can perform learning processing based on well-known methods such as the steepest descent method, neural networks, DQN (Deep Q-Network), Gaussian processes, and deep learning. Furthermore, the learning unit 130 performs learning in such a way that, for operation variables with higher reward values, they are preferentially output as more recommended operation variables. Specifically, the learning unit 130 performs reinforcement learning based on the input state data in such a way that, for operation variables with higher reward values ​​defined by a predetermined reward function, they are output as more recommended operation variables, thereby generating a control model 135. Thus, the model is updated to generate the control model 135.

[0066] In step 240, the control device 100 determines whether to terminate machine learning. If, in step 240, the control device 100 determines that learning is not to be terminated (if the answer is "No"), the control device 100 returns the process to step 210 and continues the flow. On the other hand, if, in step 240, the control device 100 determines that learning is to be terminated (if the answer is "Yes"), the control device 100 terminates the flow.

[0067] Figure 3 The following illustrates an example of a process for adjusting the manipulated variable by the control device 100 according to this embodiment. During the operational phase, the control device 100 uses the control model 135 generated through machine learning during the learning phase to perform AI control of the controlled object 20. The control device 100 then predicts the future state of the device 10 in which the controlled object 20 is installed. Based on the prediction results, the control device 100 adjusts the manipulated variable output by the control model 135.

[0068] In step 310, the control device 100 acquires status data. For example, the status data acquisition unit 110 acquires status data indicating the status of the device 10 in which the controlled object 20 is installed. As an example, the status data acquisition unit 110 acquires operating data, consumption data, and external environment data measured by sensors installed in the device 10 via a network as status data. The status data acquisition unit 110 supplies the acquired status data to the control model 135 and the prediction unit 150.

[0069] In step 320 , the control device 100 predicts the future state. For example, the prediction unit 150 predicts the future state of the device 10 based on the state data supplied from the state data acquisition unit 110 in step 310 .

[0070] For example, the upper piping of an atmospheric distillation column used in petroleum refining experiences internal corrosion, leading to thinning. In this case, the prediction unit 150 may include a prediction model that represents the relationship between the process state and the progress of thinning. Furthermore, such a prediction model can be generated by implementing pre-trained learning using process data measured at multiple locations associated with the upper piping of the atmospheric distillation column and periodic wall thickness measurement results. Furthermore, the prediction unit 150 can construct a regression equation to predict the future amount of pipe thinning. Furthermore, the prediction unit 150 can predict the time until the minimum wall thickness required to maintain the integrity of the pipe is reached (remaining life) based on the amount of thinning per unit time (thinning rate). Furthermore, the prediction unit 150 can calculate the cost of adding an inhibitor to suppress corrosion of the pipe based on at least either the future amount of pipe thinning or the remaining life. Furthermore, the prediction unit 150 can calculate the maintenance frequency based on at least either the future amount of pipe thinning or the remaining life.

[0071] In this way, the prediction unit 150 can predict the future health of the equipment 10. This health can include the remaining wall thickness of the piping of the equipment 10. Furthermore, the prediction unit 150 can calculate future maintenance costs based on the future health of the equipment 10 and predict the future maintenance costs of the equipment 10. This maintenance cost can include the cost of adding inhibitors to suppress degradation of the equipment 10. The prediction unit 150 provides the prediction results to the adjustment unit 160.

[0072] In step 330, the control device 100 acquires manipulated variable data. For example, the manipulated variable data acquisition unit 120 acquires manipulated variable data representing the manipulated variable of the controlled object 20. As an example, the manipulated variable data acquisition unit 120 acquires, from the control unit 140, data representing the manipulated variable MV(AI) output by the control model 135 when AI control is performed on the controlled object 20 as the manipulated variable data. The manipulated variable data acquisition unit 120 supplies the acquired manipulated variable data to the adjustment unit 160.

[0073] In step 340 , the control device 100 adjusts the manipulated variable MV (AI). For example, the adjuster 160 adjusts the manipulated variable MV (AI) supplied from the manipulated variable data acquirer 120 in step 330 based on the prediction result supplied from the predictor 150 in step 320 .

[0074] As an example, the adjuster 160 obtains the future health of the device 10 as a prediction result from the predictor 150. In this case, the adjuster 160 determines the operating conditions that will deteriorate the future health of the device 10 based on the relationship between the manipulated variable MV(AI) and the future health.

[0075] As an example, for an operation variable MV(AI) that causes the temperature of the fluid flowing through the pipe to be greater than or equal to a threshold, a prediction result is obtained that indicates the progression of future pipe thinning. Alternatively, for an operation variable MV(AI) that causes the temperature of the fluid flowing through the pipe to be less than a threshold, a prediction result is obtained that indicates that future pipe thinning will be suppressed. In this case, the adjustment unit 160 can determine that the primary cause of thinning is the temperature of the fluid flowing through the pipe being greater than or equal to the threshold. Furthermore, if the temperature of the fluid is greater than or equal to the threshold, this phenomenon may occur, for example, when a substance that causes corrosion to progress is generated. In this case, the adjustment unit 160 adjusts the operation variable MV(AI) to MV(AI)_adj so that the temperature of the fluid does not exceed or equal the threshold.

[0076] Similarly, for an operation variable MV(AI) that causes the velocity of the fluid flowing through the pipe to be greater than or equal to a threshold, a prediction result indicating a shortened remaining life is obtained. Furthermore, for an operation variable MV(AI) that causes the velocity of the fluid flowing through the pipe to be less than the threshold, a prediction result indicating an extended remaining life is obtained. In this case, the adjustment unit 160 can determine that the primary cause of the shortened remaining life is the velocity of the fluid flowing through the pipe being greater than or equal to the threshold. Furthermore, this phenomenon may occur in situations such as when the velocity of the fluid is greater than or equal to the threshold, causing the effects of flow-accelerated corrosion (FAC) and liquid droplet impact erosion (LDI) to become significant. In this case, the adjustment unit 160 adjusts the operation variable MV(AI) to MV(AI)_adj so that the velocity of the fluid does not exceed or equal the threshold. For example, the adjustment unit 160 may adjust the operation variable MV(AI) based on the future health of the device 10.

[0077] Furthermore, the adjuster 160 obtains the future maintenance cost of the equipment 10 as a prediction result from the predictor 150. In this case, the adjuster 160 determines an operating condition that increases the future maintenance cost of the equipment 10 based on the relationship between the operation amount MV(AI) and the future maintenance cost.

[0078] As an example, for an operation variable MV(AI) that causes the temperature of the fluid flowing through the pipe to be greater than or equal to a threshold, a prediction result is obtained that the future cost of adding a corrosion inhibitor will increase. Alternatively, for an operation variable MV(AI) that causes the temperature of the fluid flowing through the pipe to be less than the threshold, a prediction result is obtained that the future cost of adding a corrosion inhibitor will decrease. In this case, the adjustment unit 160 can determine that the main cause of the increase in the cost of adding a corrosion inhibitor is that the temperature of the fluid flowing through the pipe is greater than or equal to the threshold. In this case, the adjustment unit 160 adjusts the operation variable MV(AI) to MV(AI)_adj so that the temperature of the fluid does not exceed or equal the threshold.

[0079] Similarly, for an operation variable MV(AI) that causes the velocity of the fluid flowing through the pipe to be greater than or equal to a threshold, a prediction result is obtained that the future maintenance frequency will increase (a prediction result that maintenance costs may increase). Furthermore, for an operation variable MV(AI) that causes the velocity of the fluid flowing through the pipe to be less than a threshold, a prediction result is obtained that the future maintenance frequency will decrease (a prediction result that maintenance costs may decrease). In this case, the adjustment unit 160 can determine that the cause of the increase in maintenance costs is that the velocity of the fluid flowing through the pipe is greater than or equal to the threshold. In this case, the adjustment unit 160 adjusts the operation variable MV(AI) to MV(AI)_adj so that the velocity of the fluid does not exceed or equal the threshold. For example, the adjustment unit 160 can adjust the operation variable MV(AI) based on the future maintenance cost of the device 10. The adjustment unit 160 supplies the adjusted operation variable MV(AI)_adj to the output unit 170.

[0080] In step 350, the control device 100 outputs the adjusted manipulated variable MV(AI)_adj. For example, the output unit 170 outputs the manipulated variable MV(AI)_adj adjusted by the adjustment unit 160 in step 340 to the controlled object 20. Thus, the control device 100 controls the controlled object 20 using the manipulated variable MV(AI) adjusted based on the prediction result.

[0081] Typically, in machine learning, input data is used to determine the parameters of the learning model. There is a probability that these parameters will be found, which cannot be guaranteed in theory. Therefore, it is possible that abnormal inferred data will be output from the learning model. Therefore, the control device 100 involved in this embodiment predicts the future state of the device 10 provided with the control object 20 when performing AI control on the control object 20. Moreover, the control device 100 adjusts the manipulated variable output by the control model 135 based on the prediction result. Thus, according to the control device 100 involved in this embodiment, when it is predicted that the AI ​​control may have an adverse effect on the future state of the device 10, the manipulated variable MV(AI) of the AI ​​control is adjusted, and the controlled object 20 can be controlled using the adjusted manipulated variable MV(AI)_adj. Here, when the control model 135 is generated by machine learning, the future prediction result of the device 10 (for example, including a reward function, etc.) is also taken into account and machine learning is performed in a manner such that the control model 135 outputs the recommended manipulated variable MV(AI). However, the control device 100 according to this embodiment considers that adjustments to the future prediction results of the device 10 are not embedded in the machine learning used to generate the control model 135, and the manipulated variable MV (AI) output by the control model 135 is adjusted afterward. Thus, the control device 100 according to this embodiment can further simplify the machine learning used to generate the control model 135, and can also adjust the output of an already constructed control model 135. Furthermore, the control device 100 according to this embodiment makes the adjustment of the manipulated variable MV (AI) visible, rather than being a black box in the machine learning.

[0082] Furthermore, the control device 100 according to this embodiment predicts the future health of the equipment 10 (e.g., the remaining wall thickness of the piping) and adjusts the manipulated variable based on the health. Thus, the control device 100 according to this embodiment can, for example, control the controlled object 20 while avoiding operating conditions that favor the progression of corrosion, thereby suppressing degradation of the health of the equipment 10.

[0083] Furthermore, the control device 100 according to this embodiment predicts future maintenance costs (e.g., inhibitor addition costs) for the equipment 10 and adjusts the operating amount based on the maintenance costs. Thus, the control device 100 according to this embodiment can, for example, control the controlled object 20 while avoiding operating conditions that are likely to increase inhibitor addition costs, thereby suppressing increases in maintenance costs for the equipment 10.

[0084] Figure 4 An example of a block diagram of a device 10 provided with a control target 20 is shown together with a control device 100 according to a first modification of the present embodiment. Figure 4 In the Figure 1 Components with the same functions and structures are designated by the same reference numerals, and descriptions thereof will be omitted except for the following differences. In the control device 100 according to the first modification, machine learning constraints are learned based on the history of the adjustment manipulated variable MV(AI). The control device 100 according to the first modification further includes an adjustment history acquisition unit 410.

[0085] Adjustment history acquisition unit 410 acquires the adjustment history of manipulated variable MV(AI) from adjustment unit 160. For example, adjustment history acquisition unit 410 acquires information representing manipulated variable MV(AI) output by control model 135 and manipulated variable MV(AI)_adj adjusted by adjustment unit 160 as the adjustment history of manipulated variable MV(AI). Adjustment history acquisition unit 410 then supplies the acquired adjustment history to learning unit 130. In response, learning unit 130 further learns the constraints for machine learning based on the history of adjusted manipulated variable.

[0086] As an example, when the manipulated variable MV(AI) = 113 is adjusted to MV(AI)_adj = 100, the learning unit 130 learns a constraint condition that prohibits the control model 135 from outputting the pre-adjusted manipulated variable MV(AI) = 113. Similarly, when the manipulated variable MV(AI) = 47 is adjusted to MV(AI)_adj = 50, the learning unit 130 learns a constraint condition that prohibits the control model 135 from outputting the pre-adjusted manipulated variable MV(AI) = 47. Thus, when the manipulated variable MV(AI) is adjusted to MV(AI)_adj by the adjustment unit 160, the learning unit 130 learns the constraint condition that prohibits the control model 135 from outputting the manipulated variable MV(AI). Furthermore, by accumulating prohibited values ​​of the manipulated variable MV(AI), the learning unit 130 learns a constraint condition that sets the upper limit of the manipulated variable MV(AI) to 100 and the lower limit to 50. Thus, the constraints learned by the learning unit 130 may include at least one of an upper limit and a lower limit of the manipulated variable MV(AI) permitted to be output from the control model 135. Furthermore, the learning unit 130 relearns during the learning phase under the constraints learned in this manner, thereby regenerating the control model 135 that outputs the manipulated variable corresponding to the state of the device 10 using the state data and the manipulated variable data.

[0087] In this manner, the control device 100 according to the first modification learns the constraints for machine learning based on the history of adjustments to the manipulated variable MV(AI). Thus, the control device 100 according to the first modification can, for example, reduce the number of operating conditions that are likely to be achieved, and send them to the learning unit 130. This allows the control model 135 to be regenerated in the future so that the manipulated variable NV(AI) that is less likely to be achieved is output from the control model 135.

[0088] Figure 5 An example of a block diagram of a device 10 provided with a control target 20 is shown together with a control device 100 according to a second modification of the present embodiment. Figure 5 In the Figure 1 Components with the same functions and structures are designated by the same reference numerals, and descriptions thereof will be omitted except for the following differences. In the control device 100 according to the second modification, the production plan of the equipment 10 is considered when adjusting the manipulated variable MV(AI). The control device 100 according to the second modification further includes a plan acquisition unit 510 for acquiring the production plan of the equipment 10.

[0089] The plan acquisition unit 510 acquires the production plan for the equipment 10. For example, the plan acquisition unit 510 acquires information indicating the type, production volume, and production date and time of the products to be produced by the equipment 10 via a network. However, this is not a limitation. The plan acquisition unit 510 may acquire this production plan from an operator or from various storage devices. The plan acquisition unit 510 supplies the acquired production plan to the adjustment unit 160.

[0090] The adjustment unit 160 adjusts the manipulated variable MV(AI) based on the prediction results provided by the prediction unit 150 and the production plan provided by the plan acquisition unit 510. For example, the adjustment unit 160 decides to adjust the manipulated variable MV(AI) to the manipulated variable MV(AI)_adj based on the prediction results. In this case, the adjustment unit 160 examines the impact of the change in manipulated variable on productivity. As an example, the adjustment unit 160 decides to adjust the manipulated variable MV(AI) = 113 to the manipulated variable MV(AI)_adj = 100. In this case, the adjustment unit 160 estimates the production volume per unit time when the manipulated variable MV(AI) = 113 is changed to the manipulated variable MV(AI)_adj = 100. As a result, under control based on the adjusted manipulated variable MV(AI)_adj, it is estimated that the product specified in the acquired production plan cannot be produced at the specified production volume on the specified production date and time. On the other hand, if the manipulated variable MV(AI)_adj is changed to 105, it is estimated that the specified product can be produced at the specified production volume on the specified production date and time. In this case, the adjustment unit 160 adjusts the manipulated variable MV(AI)_adj = 100 to a value between the manipulated variable MV(AI) and the manipulated variable MV(AI)_adj, for example, MV(AI)_adj′ = 105. That is, the adjustment unit 160 adjusts the manipulated variable MV(AI) based on the forecast results within a range that satisfies the production plan. The adjustment unit 160 then supplies the adjusted manipulated variable MV(AI)_adj′ to the output unit 170. In response, the output unit 170 outputs the manipulated variable MV(AI)_adj′ adjusted by the adjustment unit 160 to the controlled object 20. Thus, the control device 100 controls the controlled object 20 using the manipulated variable MV(AI)_adj′ adjusted based on the forecast results and the production plan.

[0091] Figure 6 An example of a flow of adjusting the control amount by the control device 100 according to the second modified example of the present embodiment is shown. Figure 6 In, with Figure 3 The same processing is denoted by the same reference numerals, and description thereof will be omitted except for the following differences. This flow further includes steps 610 to 630.

[0092] In step 610, the control device 100 determines whether the production plan is satisfied. For example, the adjustment unit 160 estimates the production volume per unit time when the controlled object 20 is assigned the manipulated variable MV(AI)_adj adjusted in step 340. In this case, as an example, the adjustment unit 160 can estimate the production volume per unit time using a known relationship that represents the relationship between the manipulated variable MV and the production volume per unit time. Furthermore, the adjustment unit 160 calculates the period from the current moment to the production date and time specified in the production plan for each product specified in the production plan. The adjustment unit 160 multiplies the estimated production volume per unit time by the period until the production date and time to estimate the production volume that can be produced until the production date and time. If the predicted production volume is greater than or equal to the production volume specified in the production plan, the adjustment unit 160 determines that the production plan is satisfied (Yes). If, in step 610, it is determined that the production plan is satisfied, the control device 100 proceeds to step 350. That is, the control device 100 outputs the manipulated variable MV(AI)_adj to the controlled object 20 and controls the controlled object 20 using the manipulated variable MV(AI)_adj.

[0093] On the other hand, if the predicted production volume is lower than the production volume specified in the production plan, the adjustment unit 160 determines that the production plan is not satisfied (No). If it is determined in step 610 that the production plan is not satisfied, the adjustment unit 160 advances the process to step 620.

[0094] In step 620, the control device 100 readjusts the manipulated variable MV(AI). For example, the adjustment unit 160 uses trial and error between the manipulated variable MV(AI) and the manipulated variable MV(AI)_adj to find the manipulated variable MV(AI)_adj′ that satisfies the production plan. The adjustment unit 160 then readjusts the manipulated variable MV(AI)_adj to the found manipulated variable MV(AI)_adj′. The adjustment unit 160 supplies the readjusted manipulated variable MV(AI)_adj′ to the output unit 170.

[0095] In step 630, the control device 100 outputs the manipulated variable MV(AI)_adj′. For example, the output unit 170 outputs the manipulated variable MV(AI)_adj′ adjusted by the adjustment unit 160 in step 630 to the controlled object 20. Thus, the control device 100 controls the controlled object 20 using the manipulated variable MV(AI)_adj′ adjusted based on the prediction results and the production plan.

[0096] The above process describes in detail, for example, a case where the target manipulated variable MV is the temperature of a specific portion of the equipment 10, and the target production quantity is the amount of electricity that the equipment 10 should produce per day. For example, as a production plan, the plan acquisition unit 510 acquires the amount of electricity that the equipment 10 should produce per day for summer and winter, and for spring and autumn. The amount of electricity that should be produced per day varies depending on the supply and demand balance. For example, the amount of electricity that should be produced per day in summer and winter is 1000 Wh / day, while the amount of electricity that should be produced per day in spring and autumn is 700 Wh / day. Furthermore, the adjustment unit 160 decides to adjust the target manipulated variable MV, i.e., the temperature, from 113 degrees Celsius to 100 degrees Celsius.

[0097] In this case, the adjustment unit 160 uses a known relationship to estimate that the amount of electricity that can be produced per hour is 38Wh / hour when the temperature is changed to 100 degrees. Moreover, the adjustment unit 160 multiplies the amount of electricity that can be produced per hour by the time per day, that is, 24, and predicts that the amount of electricity that can be produced per day is 912Wh / day (=38Wh / hour×24 hours). Here, in spring and autumn, the amount of electricity that should be produced per day is 700Wh / day, and the predicted production volume (912Wh / day) is greater than or equal to the production plan specified by the production plan (700Wh / day). In this case, the adjustment unit 160 determines that the production plan is met and adjusts the object operation variable MV, that is, the temperature, from 113 degrees to 100 degrees. Accordingly, the control device 100 outputs the adjusted temperature, that is, 100 degrees, to the control object 20.

[0098] On the other hand, in summer and winter, the amount of electricity that should be produced per day is 1000Wh / day, and the predicted production volume (912Wh / day) is less than the production plan (1000Wh / day) specified by the production plan. In this case, the adjustment unit 160 determines that the production plan is not met. Furthermore, the adjustment unit 160 uses a known relationship to iterate through trial and error and estimates that the amount of electricity that can be produced per hour is 42Wh / hour when the temperature is changed to 105 degrees. Furthermore, the adjustment unit 160 predicts that the amount of electricity that can be produced per day is 1008Wh / day (=42Wh / hour × 24 hours). Therefore, the predicted production volume (1008Wh / day) is greater than or equal to the production plan (1000Wh / day) specified by the production plan. In this case, the adjustment unit 160 determines that the production plan is met and adjusts the temperature, which is the object of the manipulated variable MV, to 105 degrees. Accordingly, the control device 100 outputs the adjusted temperature, i.e., 105 degrees, to the controlled object 20.

[0099] In addition, the above description shows, as an example, a case where the adjustment unit 160 adjusts the target operating variable from 113 degrees to 100 degrees in spring and autumn when the predicted production volume is greater than or equal to the production plan specified by the production plan. However, this is not limiting. The adjustment unit 160 may further readjust the target operating variable if it determines that the production plan is met. For example, the adjustment unit 160 uses a known relationship through trial and error to estimate that the amount of electricity that can be produced per hour when the temperature is changed to 90 degrees is 30 Wh / hour. Furthermore, the adjustment unit 160 predicts that the amount of electricity that can be produced per day is 720 Wh / day (= 30 Wh / hour x 24 hours). Even with this temperature change, the predicted production volume (720 Wh / day) is greater than or equal to the production plan specified by the production plan (700 Wh / day). In this case, the adjustment unit 160 may further readjust the target operating variable MV, for example, adjusting the temperature to 90 degrees, which is lower than 100 degrees. As a result, the control device 100 can readjust the target operation amount MV so as to further suppress thinning, for example, within a range that satisfies the production plan.

[0100] Thus, the control device 100 according to the second modification takes the production plan of the equipment 10 into consideration when adjusting the manipulated variable MV(AI). Specifically, the control device 100 according to the second modification adjusts the manipulated variable MV(AI) within a range that satisfies the production plan. Thus, the control device 100 according to the second modification achieves both extending the life of the equipment 10 and ensuring adherence to the production plan.

[0101] Various embodiments of the present invention may be described with reference to flowcharts and block diagrams, where a module may represent (1) a stage of a process for performing an operation, or (2) a portion of a device having the function of performing an operation. Specific stages and portions may be implemented using dedicated circuits, programmable circuits supplied with computer-readable instructions stored on a computer-readable medium, and / or processors supplied with computer-readable instructions stored on a computer-readable medium. Dedicated circuits may include digital and / or analog hardware circuits, and may also include integrated circuits (ICs) and / or discrete circuits. Programmable circuits may include reconfigurable hardware circuits including logical AND, logical OR, logical XOR, logical NAND, logical NOR, and other logical operations, flip-flop circuits, registers, memory elements such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and the like.

[0102] The computer-readable medium may include any tangible device capable of storing instructions for execution using an appropriate device. As a result, the computer-readable medium having the instructions stored therein has a product containing executable instructions for making a means for performing the operations specified in the flowchart or block diagram. Examples of computer-readable media include electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, etc. More specific examples of computer-readable media include Floppy (registered trademark) floppy disks, floppy disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electrically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), compact disc read-only memory (CD-ROM), digital versatile disk (DVD), Blu-ray (RTM) disks, memory sticks, integrated circuit cards, etc.

[0103] Computer readable instructions may include any source code or object code described in assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or any combination of one or more programming languages ​​including object-oriented programming languages ​​such as Smalltalk, JAVA (registered trademark), C++, and procedural programming languages ​​such as the "C" programming language or similar programming languages.

[0104] Computer-readable instructions may be provided to a processor or programmable circuit of a general-purpose computer, a special-purpose computer, or other programmable data processing device via a local area network (LAN), a wide area network (WAN), or the like, and the computer-readable instructions may be executed to create a means for performing the operations specified in the flowchart or block diagram. Examples of processors include computer processors, processing units, microprocessors, digital signal processors, controllers, and microcontrollers.

[0105] Figure 7 This represents an example of a computer 9900 that can embody all or part of the various aspects of the present invention. The program installed in computer 9900 can cause computer 9900 to function as an operation associated with a device according to an embodiment of the present invention or one or more components of such a device, or to execute such operation or such one or more components, and / or can cause computer 9900 to perform a process or a stage of such a process according to an embodiment of the present invention. This program can be executed by CPU 9912 to cause computer 9900 to perform specific operations associated with some or all of the modules in the flowcharts and block diagrams described in this specification.

[0106] The computer 9900 according to this embodiment includes a CPU 9912, a RAM 9914, a graphics controller 9916, and a display device 9918, and these components are interconnected via a main controller 9910. Furthermore, the computer 9900 includes input / output units such as a communication interface 9922, a hard disk drive 9924, a DVD-ROM drive 9926, and an IC card drive, and these components are connected to the main controller 9910 via an input / output controller 9920. Furthermore, the computer includes conventional input / output units such as a ROM 9930 and a keyboard 9942, and these components are connected to the input / output controller 9920 via an input / output chip 9940.

[0107] The CPU 9912 controls each unit by executing operations according to programs stored in the ROM 9930 and the RAM 9914. The graphics controller 9916 obtains image data generated by the CPU 9912 in a frame buffer or the like or in the graphics controller itself and supplies it to the RAM 9914, and displays the image data on the display device 9918.

[0108] The communication interface 9922 communicates with other electronic devices via a network. The hard disk drive 9924 stores programs and data used by the CPU 9912 in the computer 9900. The DVD-ROM drive 9926 reads programs and data from the DVD-ROM 9901 and provides the programs and data to the hard disk drive 9924 via the RAM 9914. The IC card driver reads programs and data from an IC card and / or writes programs and data to an IC card.

[0109] The ROM 9930 stores therein a startup program and the like executed by the computer 9900 upon activation, and / or programs that depend on the hardware of the computer 9900. In addition, the input / output chip 9940 connects various input / output units to the input / output controller 9920 via a parallel port, a serial port, a keyboard port, a mouse port, and the like.

[0110] The program is provided on a computer-readable medium such as a DVD-ROM 9901 or an IC card. The program is read from the computer-readable medium and installed in a hard disk drive 9924, a RAM 9914, or a ROM 9930, also an example of a computer-readable medium, and then executed by the CPU 9912. The information processing described in the program is read by the computer 9900, and the program and the various types of hardware resources described above are coordinated. By using the computer 9900 to implement information manipulation or processing, a device or method can be constructed.

[0111] For example, when communication is performed between the computer 9900 and an external device, the CPU 9912 can execute a communication program loaded in the RAM 9914 and, based on the processing described in the communication program, issue communication processing instructions to the communication interface 9922. Under the control of the CPU 9912, the communication interface 9922 reads transmission data stored in a transmission buffer area provided in the RAM 9914, the hard disk drive 9924, the DVD-ROM 9901, or a recording medium such as an IC card, and transmits the read transmission data to the network or writes reception data received from the network to a reception buffer area provided on the recording medium.

[0112] Furthermore, the CPU 9912 can read all or a required portion of a file or database stored in an external recording medium such as the hard disk drive 9924, DVD-ROM drive 9926 (DVD-ROM 9901), or IC card into the RAM 9914, and perform various types of processing on the data in the RAM 9914. The CPU 9912 then writes the processed data back to the external recording medium.

[0113] Various types of information such as various types of programs, data, tables, and databases can be stored in a recording medium and subjected to information processing. The CPU 9912 can perform various types of processing including various types of operations, information processing, conditional judgments, conditional branches, unconditional branches, information retrieval / replacement, etc., which are recorded at any location in the present disclosure and specified by the instruction sequence of the program, on the data read from the RAM 9914, and write the results back to the RAM 9914. In addition, the CPU 9912 can retrieve information in files, databases, etc. in the recording medium. For example, in the case where a plurality of entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored in the recording medium, the CPU 9912 can retrieve an entry that specifies an attribute value of the first attribute that is consistent with the condition from the plurality of records, read the attribute value of the second attribute stored in the entry, and obtain the attribute value of the second attribute associated with the first attribute that satisfies the condition predetermined thereby.

[0114] The program or software module described above can be stored in a computer-readable medium on or near the computer 9900. In addition, a recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can be used as a computer-readable medium, thereby providing the program to the computer 9900 via the network.

[0115] The present invention has been described above using the embodiments. However, the technical scope of the present invention is not limited to the scope described in the above embodiments. Those skilled in the art will appreciate that various modifications or improvements may be made to the above embodiments. As is clear from the claims, embodiments incorporating such modifications or improvements are also within the technical scope of the present invention.

[0116] Regarding the order in which actions, sequences, steps, and stages, etc., of the apparatuses, systems, programs, and methods described in the claims, specifications, and drawings are executed, it should be noted that unless otherwise expressly indicated as "before," "before," or the like, and unless the output of a previous process is used in a subsequent process, the execution order may be any order. Even if the flow of actions in the claims, specifications, and drawings is described using the phrases "first," "next," or the like for convenience, it does not necessarily mean that the actions must be executed in that order.

[0117] Description of the label

[0118] 10 Equipment

[0119] 20 Control Objects

[0120] 100 Control Device

[0121] 110 Status data acquisition unit

[0122] 120 Operation volume data acquisition unit

[0123] 130 Learning Department

[0124] 135 Control Model

[0125] 140 Control Department

[0126] 150 Forecasting Department

[0127] 160 Adjustment Department

[0128] 170 Output

[0129] 410 Adjustment of Resume Acquisition Department

[0130] 510 Program Acquisition Department

[0131] 9900 Computer

[0132] 9901 DVD-ROM

[0133] 9910 Master Controller

[0134] 9912 CPU

[0135] 9914 RAM

[0136] 9916 Graphics Controller

[0137] 9918 Display Device

[0138] 9920 Input / Output Controller

[0139] 9922 Communication Interface

[0140] 9924 Hard Drive

[0141] 9926 DVD drive

[0142] 9930 ROM

[0143] 9940 Input / Output Chip

[0144] 9942 Keyboard

Claims

1. A control device, wherein: The control device has: a control unit that controls the control object using a control model obtained by machine learning in a manner that outputs an operation amount of the control object in accordance with a state of a device provided with the control object; a prediction unit configured to predict a future state of the device; an adjusting unit configured to adjust the operation amount based on the prediction result; an output unit that outputs the adjusted operation amount to the controlled object; and a learning unit that generates the control model through machine learning, The learning unit further learns the constraint conditions of the machine learning based on the history of adjusting the operation amount. The learning unit further learns the constraint condition so as to prohibit the manipulated variable before adjustment from being output from the control model.

2. The control device according to claim 1, wherein: The prediction unit predicts the future health of the device, The adjustment unit adjusts the operation amount according to the healthiness.

3. The control device according to claim 2, wherein: The soundness includes the remaining wall thickness of the piping of the equipment.

4. The control device according to claim 1, wherein: The prediction unit predicts the future maintenance cost of the equipment, The adjustment unit adjusts the operation amount according to the maintenance cost.

5. The control device according to claim 2, wherein: The prediction unit predicts the future maintenance cost of the equipment, The adjustment unit adjusts the operation amount according to the maintenance cost.

6. The control device according to claim 3, wherein: The prediction unit predicts the future maintenance cost of the equipment, The adjustment unit adjusts the operation amount according to the maintenance cost.

7. The control device according to claim 4, wherein: The maintenance cost includes the cost associated with the addition of an inhibitor to suppress degradation of the equipment.

8. The control device according to claim 5, wherein: The maintenance cost includes the cost associated with the addition of an inhibitor to suppress degradation of the equipment.

9. The control device according to claim 6, wherein: The maintenance cost includes the cost associated with the addition of an inhibitor to suppress degradation of the equipment.

10. The control device according to claim 1, wherein: The learning unit generates the control model by performing reinforcement learning in response to input of state data indicating the state of the device so as to output a more recommended operation amount for an operation amount having a higher reward value determined by a predetermined reward function.

11. The control device according to claim 1, wherein: The restriction condition includes at least one of an upper limit value and a lower limit value of the operation amount.

12. The control device according to any one of claims 1 to 11, wherein: The control device further includes a plan acquisition unit that acquires a production plan of the equipment. The adjustment unit adjusts the operation amount based on a prediction result and a production plan.

13. A control method, wherein: The control method has the following steps: Controlling the control object using a control model obtained by machine learning in a manner that outputs an operation amount of the control object in accordance with a state of a device provided with the control object; predicting the future state of the device; adjusting the operation amount based on the prediction result; outputting the adjusted operation amount to the control object; and The control model is generated by machine learning, Furthermore, based on the history of adjusting the operation amount, the constraint conditions of the machine learning are learned. The restriction condition is also learned in such a manner as to prohibit the manipulated variable before adjustment from being output from the control model.

14. A recording medium having a control program recorded thereon, wherein: The control program is executed by a computer, causing the computer to function as the following functional unit: a control unit that controls the control object using a control model obtained by machine learning in a manner that outputs an operation amount of the control object in accordance with a state of a device provided with the control object; a prediction unit configured to predict a future state of the device; an adjusting unit configured to adjust the operation amount based on the prediction result; an output unit that outputs the adjusted operation amount to the controlled object; and a learning unit that generates the control model through machine learning, The learning unit further learns the constraint conditions of the machine learning based on the history of adjusting the operation amount. The learning unit further learns the constraint condition so as to prohibit the manipulated variable before adjustment from being output from the control model.

Images