JP7655478B2 - MODEL PREDICTIVE CONTROL SYSTEM, INFORMATION PROCESSING DEVICE, PROGRAM, AND MODEL PREDICTIVE CONTROL METHOD - Google Patents

Description

The present invention relates to a model predictive control system, an information processing device, a program, and a model predictive control method.

Conventionally, model predictive control (MPC) using a mathematical model of the process to be controlled (hereinafter also referred to as "process model") has been known as a process control method. In addition, to address the issue that the accuracy of model predictive control can decrease when the characteristics of the process change due to factors such as aging of equipment, a method of periodically updating the process model is known. For example, Patent Document 1 discloses a model predictive control method using online model identification, which discloses a technology for model re-identification in which a process dynamic model to be used is derived each time a model predictive control calculation is performed based on actual operation data.

JP 2011-198327 A

In conventional technology, model re-identification is performed periodically. However, in a configuration in which model re-identification is performed every cycle, the accuracy of model predictive control may not necessarily improve. Therefore, there is room for improvement in technology related to model predictive control of processes.

In light of these circumstances, the objective of the present invention is to improve technology related to model predictive control of processes.

A model predictive control system according to an embodiment of the present invention includes:
a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A model predictive control system comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The update means executes the update process except when the decision means decides not to execute the update process.

An information processing device according to an embodiment of the present invention includes:
a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
An information processing device comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The update means executes the update process except when the decision means decides not to execute the update process.

An information processing device according to an embodiment of the present invention includes:
A model predictive control system is provided that includes a storage means for storing a process model, a process control means for performing model predictive control of a process using the process model, a storage means for storing detection values of a control output and a control input of the process as time-series data, an update means capable of executing an update process for updating a model parameter of the process model based on the time-series data stored during a learning period, and a decision means for deciding whether or not to execute the update process, and the update means executes the update process except when it is determined by the decision means not to execute the update process.
The system includes at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means.

A program according to an embodiment of the present invention includes:
An information processing device,
a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A program that functions as
causing the information processing device to further function as a determination unit for determining whether or not to execute the update process;
The update means executes the update process except when the decision means decides not to execute the update process.

A program according to an embodiment of the present invention includes:
a process control means for performing model predictive control of a process using the process model; a storage means for storing a process model; a storage means for storing detection values of a control output and a control input of the process as time-series data; an update means for executing an update process for updating a model parameter of the process model based on the time-series data accumulated during a learning period; and a decision means for deciding whether or not to execute the update process. In the model predictive control system, the update means executes the update process by selecting one of the information processing devices, except when the decision means decides not to execute the update process.
The device functions as at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means.

A model predictive control method according to an embodiment of the present invention includes:
A model predictive control method for a process executed by an information processing device, comprising:
storing the process model;
performing model predictive control of a process using the process model;
accumulating detection values of control outputs and control inputs of the process as time series data;
determining whether to execute an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
and performing the update process unless it is determined not to perform the update process.

One embodiment of the present invention improves technology related to model predictive control of processes.

1 is a diagram showing a schematic configuration of a model predictive control system according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an update period in model predictive control. 1 is a block diagram showing a schematic configuration of an information processing device; FIG. 1 is a diagram showing time-series data stored in an information processing device. 11 is a flowchart showing a first operation of the information processing device. 13 is a flowchart showing a second operation of the information processing device.

The following describes an embodiment of the present invention.

(Configuration of model predictive control system)
A model predictive control system 1 according to an embodiment of the present invention will be described with reference to Fig. 1. The model predictive control system 1 is a system used to control a process in any plant, such as a water treatment plant. The model predictive control system 1 includes an information processing device 10 and one or more plant facilities 20. Although Fig. 1 illustrates one information processing device 10 and three plant facilities 20, the model predictive control system 1 can include any number of information processing devices 10 and plant facilities 20.

The plant equipment 20 is any equipment used to carry out a process in a plant. For example, the plant equipment 20 used to carry out a water purification process in a water treatment plant is equipment that includes field equipment such as a blower that blows air to the treated water (e.g., wastewater), a control device such as a PLC (Programmable Logic Controller) that controls the field equipment, a sensor (detection means) that detects process data such as the nitric acid concentration and/or ammonia concentration in the treated water, and an operation amount such as the blowing amount of air blown to the treated water or the driving power of the blower, and a communication device that communicates with the information processing device 10 via a network N such as the Internet or a LAN (Local Area Network). However, the configuration of the plant equipment 20 is not limited to this example, and it is also possible to further include a storage device that temporarily stores the detection value of the sensor.

The information processing device 10 is installed, for example, in a central control room of the plant, but is not limited to this and can be installed anywhere. The information processing device 10 can communicate with plant equipment 20 installed in the plant via the network N.

The information processing device 10 is a device that performs model predictive control of any process to be controlled. Specifically, the information processing device 10 predicts future control output using a process model of the process to be controlled for each control cycle, and determines a control input so as to bring the actual control output closer to a target value. For example, when a water purification process in a water treatment plant is to be controlled, process data such as the nitric acid concentration and/or ammonia concentration in the treated water can be the control output, and an operating amount such as the volume of air blown to the treated water or the driving power of a blower can be the control input. However, the process to be controlled, and the control output and control input of the process are not limited to this example and can be determined arbitrarily.

In addition, the information processing device 10 can execute a process (hereinafter also referred to as "update process") to update the model parameters of the process model for each update period while model predictive control is being performed. Specifically, while model predictive control is being performed, the information processing device 10 accumulates the detection values of the control output and control input of the process as time series data for each control period. The information processing device 10 can execute an update process to update the model parameters of the process model based on the time series data (hereinafter also referred to as "learning data") accumulated during a learning period described below.

In this embodiment, the update process may be performed during a processing period described later. For example, FIG. 2 is a diagram showing the mth and m+1th update cycles. Each update cycle includes a learning period and a processing period following the learning period. The learning period is a period for accumulating learning data used in the model parameter update process. The processing period is a period for determining whether or not to perform the model parameter update process and for performing the update process according to the result of the determination. In FIG. 2, T indicates the length of the update cycle, Ta indicates the length of the learning period, and Tb indicates the length of the processing period. Note that the length of the control cycle of the model predictive control is sufficiently short compared with the lengths T, Ta, and Tb of the update cycle, learning period, and processing period, respectively. For example, the model parameters may be updated during the processing period of the mth update cycle based on the learning data accumulated during the learning period of the mth update cycle. Then, when the model parameters are updated, the model predictive control is performed during the m+1th update cycle using the updated model parameters. On the other hand, when the model parameters are not updated, the model predictive control is performed during the m+1th update cycle using the model parameters with the same values as those during the mth update cycle.

In this embodiment, the information processing device 10 determines whether to perform an update process for each update period, and performs the update process except when it is determined not to perform the update process. With this configuration, it is automatically determined whether to perform an update process for the model parameters. Therefore, by appropriately determining whether to perform an update process, it is possible to update the model parameters while avoiding inappropriate situations, such as when an abnormality occurs in a sensor that detects the control output and control input, thereby improving technology related to model predictive control of processes.

(Hardware configuration of information processing device)
The hardware configuration of the information processing device 10 will be described with reference to Fig. 3. The information processing device 10 includes a communication unit 11, a storage unit 12, and a control unit 13.

The communication unit 11 is one or more communication interfaces that communicate with external devices wirelessly or via wires. In this embodiment, the communication unit 11 includes a communication interface that communicates with the plant equipment 20 via the network N. The communication unit 11 also includes communication interfaces that communicate with each of the display device 30 and the input device 40. Here, the display device 30 is any display, such as a liquid crystal display or an OEL (Organic Electro-luminescence) display. The input device 40 is any input interface that accepts operations by a user, such as a keyboard or a mouse.

The storage unit 12 is one or more memories. In this embodiment, the "memory" is, for example, a semiconductor memory, a magnetic memory, or an optical memory, but is not limited to these and can be any memory. Each memory included in the storage unit 12 functions as, for example, a main memory device, an auxiliary memory device, or a cache memory. The storage unit 12 is, for example, built into the information processing device 10, but can also be configured to be connected externally to the information processing device 10 via any interface.

The control unit 13 includes one or more processors. In this embodiment, the "processor" may be a general-purpose processor, a dedicated processor specialized for a particular process, or the like, but is not limited to these and may be any processor. The control unit 13 controls the operation of the entire information processing device 10. The operation of the information processing device 10 will be described in detail later.

(Software configuration of information processing device)
The software configuration of the information processing device 10 will be described with reference to Fig. 3. One or more programs used to control the operation of the information processing device 10 are stored in the storage unit 12. When the one or more programs are loaded by the control unit 13, they cause the control unit 13 to function as a storage unit 131, a process control unit 132, an accumulation unit 133, an update unit 134, and a determination unit 135.

The following provides an overview of each of the means 131 to 135. The storage means 131 is a means for storing a process model of the process to be controlled in the storage unit 12. The process model is a mathematical model having one or more model parameters. The process model further includes information on the initial values of each model parameter. The process model can be determined in advance, for example, by experiment or simulation.

The process control means 132 is a means for implementing model predictive control of the process using the process model stored in the memory unit 12. Specifically, the process control means 132 predicts future control output (e.g., process data such as nitric acid concentration and/or ammonia concentration) for each control cycle based on the process model stored in the memory unit 12, determines control input (e.g., operation amount such as air volume or driving power of the blower) so as to bring the actual control output closer to the target value, and inputs the determined command value of the control input to the plant equipment 20. The target value of the control output can be changed at any timing, either in response to a user operation or automatically.

The accumulation means 133 is a means for accumulating the target value of the process control output, the detected value of the control output, and the detected value of the control input in the memory unit 12 as time series data for each control cycle. For example, as shown in FIG. 4, a combination of time data, the target value of the control output, the detected value of the control output, and the detected value of the control input is accumulated in the memory unit 12 as time series data. Note that the detected values of the control output and the control input are numerical values detected by a sensor in the plant equipment 20. In addition, the detected value of the control input by the sensor in the plant equipment 20 does not necessarily have to match the command value of the control input input to the plant equipment 20 by the process control means 132 described above.

The update means 134 is a means for executing an update process that updates the model parameters of the process model based on the time series data (i.e., learning data) accumulated in the memory unit 12 during the learning period. As described below, the update means 134 executes the update process except when the decision means 135 decides not to execute the update process.

The decision means 135 is a means for deciding whether or not to execute an update process for each update period. Here, any algorithm can be used to decide whether or not to execute an update process. Specific examples of such algorithms will be described later.

In addition to the above-mentioned means 131 to 135, the control unit 13 can also be configured to function as a means for performing any processing that can generally be performed by a processor (for example, a means for performing general arithmetic processing and timing processing, etc.).

(Operation of information processing device)
The operation of the information processing device 10 will be described with reference to Figures 5 and 6. Figure 5 is a flow chart showing a first operation for implementing model predictive control of a process.

Step S100: The storage means 131 stores the process model of the process to be controlled in the storage unit 12.

Step S101: The control unit 13 performs initial settings for the model predictive control. Specifically, the control unit 13 sets a variable i to 1. The variable i is a variable that indicates which update period the current time, calculated from the start of the model predictive control, belongs to. If the current time, calculated from the start of the model predictive control, is t and the length of the update period is T, then when (i-1)T≦t<iT, the current time t belongs to the i-th update period. In addition, the control unit 13 sets the model parameters of the process model to initial values.

Step S102: The process control means 132 predicts future control output based on the process model and determines a control input so as to bring the actual control output closer to the target value. The process control means 132 inputs the determined command value of the control input to the plant equipment 20 via the communication unit 11. Note that the process control means 132 can change the target value of the control output at any timing, either automatically or in response to a user operation, as described above.

Step S103: When the accumulation means 133 acquires the detected values of the control output and control input of the process from the plant equipment 20 via the communication unit 11, it accumulates a combination of time data, the target value of the control output, the detected value of the control output, and the detected value of the control input as time series data in the memory unit 12. After that, the operation returns to step S102. Steps S102 and S103 are repeatedly executed for each control period.

Figure 6 is a flowchart showing the second operation for executing the model parameter update process. The second operation is performed within the processing period of each update period. In addition, the second operation is performed in parallel with the process of repeating steps S102 and S103 in the first operation described above.

Step S200: The decision means 135 decides whether or not to execute a model parameter update process based on the learning data accumulated during the learning period of the i-th update cycle. A specific algorithm for deciding whether or not to execute an update process will be described later. If it is decided not to execute an update process (step S200-No), the operation proceeds to step S204. On the other hand, except when it is decided not to execute an update process (step S200-Yes), the operation proceeds to step S201.

Step S201: The update means 134 estimates model parameters based on the learning data accumulated during the learning period of the i-th update cycle. Specifically, the update means 134 searches for estimated values of the model parameters by machine learning using the learning data.

Step S202: The control unit 13 waits until the end of the i-th update cycle. When the i-th update cycle ends (step S202-Yes), the operation proceeds to step S203.

Step S203: The update means 134 updates the model parameters with the estimated values found in step S201. Then, the operation proceeds to step S205.

Step S204: The control unit 13 waits until the end of the i-th update cycle. When the i-th update cycle ends (step S204-Yes), the operation proceeds to step S205.

Step S205: The control unit 13 increments the variable i. Then, this operation ends.

(Operation of the Determining Means)
The operation of the decision means 135 for deciding whether or not to execute the update process of the model parameters will be specifically described. In this embodiment, there are a number of algorithms by which the decision means 135 decides not to execute the update process.

First, the first algorithm will be described. When the state in which the detection value of the control output or control input approximately matches the first reference value continues for a predetermined reference time during the learning period, the decision means 135 determines that there is an abnormality in the learning data and decides not to execute the update process. Here, the first reference value is the detection value of the control output or control input at any reference time during the learning period, or the upper or lower detection limit value of the sensor that detects the control output or control input. In addition, the reference time may be determined as an absolute value or a relative value (for example, n% of the length Ta of the learning period, etc.).

In this embodiment, two values being "approximately the same" means that the two values are close enough to be considered identical in practice. For example, if the absolute value of the difference between the detected value of the control output or control input and the first reference value is less than a predetermined threshold, it can be determined that the detected value is approximately the same as the first reference value. The threshold may be determined as an absolute value, or as a relative value (e.g., n% of the first reference value).

Here, in a situation different from the normal operation of the plant, an abnormality may occur in the detection value of the control output or control input (i.e., the learning data). For example, during calibration work of the plant equipment 20, the detection value may not change and may be a constant value. Also, for example, if a sensor breaks down, or if the amount of the object to be treated in the process (for example, the inflow amount of treated water in a water purification process) changes significantly compared to normal operation, the detection value may not change from the upper detection limit or the lower detection limit and may be a constant value. In contrast, according to the first algorithm, it is possible to update the model parameters while avoiding a situation where there is an abnormality in the learning data.

The second algorithm will be described. When the variance or standard deviation of the detection value of the control output or control input of the process is less than a predetermined reference value during a reference period that is at least a part of the learning period, the decision means 135 determines that there is an abnormality in the learning data and decides not to execute the update process. Here, the length of the reference period is, for example, determined in advance. When the length of the reference period is shorter than the length Ta of the learning period, the decision means 135 shifts the start position of the reference period in the learning period and sequentially determines whether or not the variance or standard deviation of the detection value of the control output or control input of the process is less than a predetermined reference value. According to the second algorithm, as with the first algorithm described above, it is possible to update the model parameters while avoiding a situation in which there is an abnormality in the learning data.

The third algorithm will now be described. If the total time during the learning period during which the detection value of the control output or control input of the process is approximately equal to a second reference value described later is equal to or greater than a predetermined reference time, the decision means 135 determines that there is an abnormality in the learning data and decides not to execute the update process. Here, the second reference value is the upper or lower detection limit value of the sensor that detects the control output or control input. The reference time may be determined as an absolute value or a relative value (for example, n% of the length Ta of the learning period, etc.).

Here, in a situation different from the normal operation of the plant, as described above, the detection value of the control output or control input may not change from the upper or lower detection limit. However, in reality, even in such a situation, the detection value may momentarily deviate from the upper or lower detection limit due to, for example, the influence of disturbances. In contrast, according to the third algorithm, even if the detection value momentarily deviates from the upper or lower detection limit, it is possible to update the model parameters while avoiding a situation in which there is an abnormality in the learning data.

The fourth algorithm will be described. When the detected value of the control output of the process is approximately equal to the target value over the learning period, the decision means 135 determines that the model predictive control is being stably performed and decides not to execute the update process. According to the fourth algorithm, when the control output of the process accurately tracks the target value, the model parameters are not updated, so that overfitting in machine learning that searches for estimated values of the model parameters can be mitigated or prevented.

The fifth algorithm will be described. When the target value of the control output of the process changes during the learning period, if the change in the target value of the control output of the process is less than a predetermined reference value, and the detection value of the control output of the process substantially matches the target value over the learning period excluding the tracking period after the change in the target value, the determining means 135 determines that the model predictive control is being stably performed and determines not to execute the update process. Here, the reference value may be determined as an absolute value or a relative value (e.g., n% of the target value or detection value of the control output). The tracking period is a period during which the control output of the process tracks the target value after the change, and can be determined by any method. For example, the period from the time the target value is changed until a predetermined time (e.g., 4 hours) has elapsed, or the period during which the numerical derivative of the moving average of the detection value of the control output continues to be equal to or greater than a predetermined reference value, can be adopted as the tracking period. According to the fifth algorithm, if the control output of the process accurately tracks before and after changes in the target value, the model parameters are not updated, making it possible to mitigate or prevent overfitting in machine learning that searches for estimated values of the model parameters.

The sixth algorithm will be described. When the target value of the control output of the process changes during the learning period, if an evaluation value indicating the tracking of the control output of the process to the changed target value (for example, the tracking time or the numerical derivative of the moving average of the detection value of the control output) is within a predetermined reference range, the decision means 135 determines that the model predictive control is being stably performed and decides not to execute the update process. Here, the "reference range" may be a range that specifies only the upper limit of the evaluation value, or may be a range that specifies the upper and lower limits of the evaluation value. According to the sixth algorithm, if the control output of the process quickly tracks the change in the target value, the model parameters are not updated, so that overfitting in machine learning that searches for estimated values of the model parameters can be mitigated or prevented.

The decision means 135 decides to execute the update process except when it is determined not to execute the update process by at least one of the first to sixth algorithms described above.

Although the present invention has been described based on the drawings and examples, it should be noted that a person skilled in the art would easily be able to make various modifications and modifications based on the present disclosure. Therefore, it should be noted that these modifications and modifications are included in the scope of the present invention. For example, the functions included in each means, step, etc. can be rearranged so as not to cause logical inconsistencies, and multiple means, steps, etc. can be combined into one or divided.

For example, the model predictive control system 1 may be configured to include multiple information processing devices 10. It may also be configured such that each component or each means of one information processing device 10 according to the embodiment is distributed across multiple devices that can communicate with each other.

In the above-described embodiment, an example of the operation of the information processing device 10 has been described with reference to FIG. 5 and FIG. 6. However, a configuration is also possible in which some steps included in the above-described operation, or some operations included in one step, are omitted as long as it is not logically inconsistent. Also, a configuration is also possible in which the order of multiple steps included in the above-described operation is changed as long as it is not logically inconsistent.

In the above embodiment, each of the means realized by the control unit 13 has been described as a software configuration, but at least some of these means may be a concept that includes software resources and/or hardware resources. For example, the storage means 131 and the accumulation means 133 may include memory.

Furthermore, a general-purpose information processing device such as a computer can be used to function as the information processing device 10 according to the embodiment described above. The information processing device can be realized by storing a program describing the processing contents for realizing each function of the information processing device 10 according to the embodiment in the memory of the information processing device, and reading and executing the program by the processor of the information processing device.

Reference Signs List 1 Model predictive control system 10 Information processing device 11 Communication unit 12 Storage unit 13 Control unit 131 Storage means 132 Process control means 133 Accumulation means 134 Update means 135 Decision means 20 Plant equipment 30 Display device 40 Input device N Network

Claims (18)

a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A model predictive control system comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The determining means is configured to determine whether or not there is an abnormality in the time series data,
When the decision means determines that there is an abnormality in the time-series data, it decides not to execute the update process;
The update means executes the update process except when the decision means decides not to execute the update process ;
The decision means decides not to execute the update process based on any one of the following (1) to (3) .
(1) the determination means determines not to execute the update process when a state in which a detection value of a control output or a control input of the process substantially matches a first reference value continues for a reference time during the learning period;
The first reference value is a detected value of a control output or control input of the process, or an upper detection limit or lower detection limit value of a sensor that detects the control output or control input of the process, at a reference time within the learning period.
(2) The decision means decides not to execute the update process if the variance or standard deviation of the detection value of the control output or control input of the process is less than a reference value during a reference period, which is at least a part of the learning period.
(3) the decision means decides not to execute the update process when a total time during which the detection value of the control output or the control input of the process substantially matches a second reference value during the learning period is equal to or longer than a reference time;
The second reference value is an upper or lower detection limit value of a sensor that detects a control output or a control input of the process. a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A model predictive control system comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The update means executes the update process except when the decision means decides not to execute the update process;
the decision means decides not to execute the update process if, when the target value of the control output of the process changes within the learning period, the amount of change in the target value of the control output of the process is less than a reference value and the detection value of the control output of the process substantially matches the target value over a period of the learning period excluding a tracking period after the change in the target value. a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A model predictive control system comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The update means executes the update process except when the decision means decides not to execute the update process;
the decision means decides not to execute the update process when, in a case where a target value of a control output of the process has changed within the learning period, an evaluation value indicating the tracking of the control output of the process to the changed target value is within a predetermined reference range. a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
An information processing device comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The determining means is configured to determine whether or not there is an abnormality in the time series data,
When the decision means determines that there is an abnormality in the time-series data, it decides not to execute the update process;
The update means executes the update process except when the decision means decides not to execute the update process ;
The information processing device, wherein the decision means decides not to execute the update process based on any one of the following (1) to (3) .
(1) the determination means determines not to execute the update process when a state in which a detection value of a control output or a control input of the process substantially matches a first reference value continues for a reference time during the learning period;
The first reference value is a detected value of a control output or control input of the process, or an upper detection limit or lower detection limit value of a sensor that detects the control output or control input of the process, at a reference time within the learning period.
(2) The decision means decides not to execute the update process if the variance or standard deviation of the detection value of the control output or control input of the process is less than a reference value during a reference period, which is at least a part of the learning period.
(3) the decision means decides not to execute the update process when a total time during which the detection value of the control output or the control input of the process substantially matches a second reference value during the learning period is equal to or longer than a reference time;
The second reference value is an upper or lower detection limit value of a sensor that detects a control output or a control input of the process. a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A model predictive control system comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The update means executes the update process except when the decision means decides not to execute the update process;
The decision means decides not to execute the update process if, when the target value of the control output of the process changes within the learning period, the amount of change in the target value of the control output of the process is less than a reference value, and the detection value of the control output of the process approximately matches the target value over a period of the learning period excluding a tracking period after the change in the target value. a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A model predictive control system comprising:
A determination unit for determining whether or not to execute the update process is further provided,
The update means executes the update process except when the decision means decides not to execute the update process;
The decision means, when a target value of the control output of the process changes within the learning period, decides not to execute the update process if an evaluation value indicating the tracking of the control output of the process to the changed target value is within a predetermined reference range. a process control means for performing model predictive control of a process using the process model; a storage means for storing a process model; a storage means for storing detection values of a control output and a control input of the process as time series data; an update means capable of executing an update process for updating a model parameter of the process model based on the time series data stored during a learning period; and a decision means for deciding whether or not to execute the update process, wherein the decision means is configured to determine the presence or absence of an abnormality in the time series data, and when the decision means determines that there is an abnormality in the time series data, the decision means decides not to execute the update process, and the update means executes the update process except when it has been determined by the decision means not to execute the update process,
The method includes at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means ,
The information processing device, wherein the decision means decides not to execute the update process based on any one of the following (1) to (3) .
(1) the determination means determines not to execute the update process when a state in which a detection value of a control output or a control input of the process substantially matches a first reference value continues for a reference time during the learning period;
The first reference value is a detected value of a control output or control input of the process, or an upper detection limit or lower detection limit value of a sensor that detects the control output or control input of the process, at a reference time within the learning period.
(2) The decision means decides not to execute the update process if the variance or standard deviation of the detection value of the control output or control input of the process is less than a reference value during a reference period, which is at least a part of the learning period.
(3) the decision means decides not to execute the update process when a total time during which the detection value of the control output or the control input of the process substantially matches a second reference value during the learning period is equal to or longer than a reference time;
The second reference value is an upper or lower detection limit value of a sensor that detects a control output or a control input of the process. a storage means for storing a process model, a process control means for performing model predictive control of a process using the process model, a storage means for storing detection values of a control output and a control input of the process as time series data, a storage means for storing detection values of a control output and a control input of the process as time series data stored during a learning period, an update means capable of executing an update process for updating a model parameter of the process model based on the time series data stored during a learning period, and a decision means for deciding whether to execute the update process, wherein the update means executes the update process except when it has been determined by the decision means not to execute the update process, and the decision means decides not to execute the update process when a target value of a control output of the process has changed within the learning period, an amount of change in the target value of the control output of the process is less than a reference value, and the detection value of the control output of the process substantially matches the target value over a period of the learning period excluding a tracking period after the change in the target value,
An information processing apparatus comprising at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means. a process control means for performing model predictive control of a process using the process model; a storage means for storing a process model; a storage means for storing detection values of a control output and a control input of the process as time series data; an update means capable of executing an update process for updating a model parameter of the process model based on the time series data accumulated during a learning period; and a decision means for deciding whether to execute the update process, wherein the update means executes the update process except when it is determined by the decision means not to execute the update process, and the decision means decides not to execute the update process when a target value of a control output of the process has changed during the learning period and an evaluation value indicating the tracking ability of the control output of the process to the changed target value is within a predetermined reference range,
An information processing apparatus comprising at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means. An information processing device,
a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A program that functions as
causing the information processing device to further function as a determination unit for determining whether or not to execute the update process;
The determining means is configured to determine whether or not there is an abnormality in the time series data,
When the decision means determines that there is an abnormality in the time-series data, it decides not to execute the update process;
The update means executes the update process except when the decision means decides not to execute the update process ;
The decision means decides not to execute the update process based on any one of the following (1) to (3) .
(1) the determination means determines not to execute the update process when a state in which a detection value of a control output or a control input of the process substantially matches a first reference value continues for a reference time during the learning period;
The first reference value is a detected value of a control output or control input of the process, or an upper detection limit or lower detection limit value of a sensor that detects the control output or control input of the process, at a reference time within the learning period.
(2) The decision means decides not to execute the update process if the variance or standard deviation of the detection value of the control output or control input of the process is less than a reference value during a reference period, which is at least a part of the learning period.
(3) the decision means decides not to execute the update process when a total time during which the detection value of the control output or the control input of the process substantially matches a second reference value during the learning period is equal to or longer than a reference time;
The second reference value is an upper or lower detection limit value of a sensor that detects a control output or a control input of the process. An information processing device,
a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A program that functions as
causing the information processing device to further function as a determination unit for determining whether or not to execute the update process;
The update means executes the update process except when the decision means decides not to execute the update process;
The decision means decides not to execute the update process if, when the target value of the control output of the process changes within the learning period, the amount of change in the target value of the control output of the process is less than a reference value and the detection value of the control output of the process approximately matches the target value over a period of the learning period excluding a tracking period after the change in the target value. An information processing device,
a storage means for storing a process model;
a process control means for performing model predictive control of a process using the process model;
a storage means for storing detected values of a control output and a control input of the process as time series data;
an update means for executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
A program that functions as
The information processing device further functions as a determination unit that determines whether or not to execute the update process;
The update means executes the update process except when the decision means decides not to execute the update process;
The decision means, when a target value of the control output of the process changes within the learning period, decides not to execute the update process if an evaluation value indicating the tracking of the control output of the process to the changed target value is within a predetermined reference range. a process control unit that performs model predictive control of a process using the process model; a storage unit that stores a process model; a storage unit that stores detection values of a control output and a control input of the process as time series data; an update unit that is capable of executing an update process for updating a model parameter of the process model based on the time series data stored during a learning period; and a decision unit that decides whether or not to execute the update process, wherein the decision unit is configured to determine the presence or absence of an abnormality in the time series data, and when the decision unit determines that there is an abnormality in the time series data, the decision unit decides not to execute the update process, and the update unit executes the update process by selecting one of the information processing units among the plurality of information processing units, except when the decision unit has determined not to execute the update process.
functioning as at least the determining means among the storage means, the process control means, the accumulation means, the update means, and the determining means ;
The decision means decides not to execute the update process based on any one of the following (1) to (3) .
(1) the determination means determines not to execute the update process when a state in which a detection value of a control output or a control input of the process substantially matches a first reference value continues for a reference time during the learning period;
The first reference value is a detected value of a control output or control input of the process, or an upper detection limit or lower detection limit value of a sensor that detects the control output or control input of the process, at a reference time within the learning period.
(2) The decision means decides not to execute the update process if the variance or standard deviation of the detection value of the control output or control input of the process is less than a reference value during a reference period, which is at least a part of the learning period.
(3) the decision means decides not to execute the update process when a total time during which the detection value of the control output or the control input of the process substantially matches a second reference value during the learning period is equal to or longer than a reference time;
The second reference value is an upper or lower detection limit value of a sensor that detects a control output or a control input of the process. a process control unit that performs model predictive control of a process using the process model; a storage unit that stores a process model; a storage unit that stores detection values of a control output and a control input of the process as time-series data; an update unit that is capable of executing an update process for updating a model parameter of the process model based on the time-series data stored during a learning period; and a decision unit that decides whether to execute the update process, wherein the update unit executes the update process except when it is determined by the decision unit not to execute the update process, and the decision unit decides not to execute the update process when a target value of the control output of the process changes within the learning period, an amount of change in the target value of the control output of the process is less than a reference value, and the detection value of the control output of the process substantially matches the target value over a period of the learning period excluding a tracking period after the change in the target value,
A program that causes the device to function as at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means. a process control unit that performs model predictive control of a process using the process model; a storage unit that stores a process model; a storage unit that stores detection values of a control output and a control input of the process as time-series data; an update unit that is capable of executing an update process for updating a model parameter of the process model based on the time-series data stored during a learning period; and a decision unit that decides whether to execute the update process, wherein the update unit executes the update process except when it is determined by the decision unit not to execute the update process, and the decision unit decides not to execute the update process when a target value of a control output of the process has changed during the learning period and an evaluation value indicating the tracking ability of the control output of the process to the changed target value is within a predetermined reference range,
A program that causes the device to function as at least one of the storage means, the process control means, the accumulation means, the update means, and the determination means. A model predictive control method for a process executed by an information processing device, comprising:
storing the process model;
performing model predictive control of a process using the process model;
accumulating detection values of control outputs and control inputs of the process as time series data;
determining whether or not there is an abnormality in the time series data;
determining whether or not to execute an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period, depending on the presence or absence of the abnormality;
executing the update process, except when it is determined that the abnormality exists and therefore it is determined that the update process is not to be executed;
Including,
A model predictive control method comprising: determining not to execute the update process based on any one of the following (1) to (3) :
(1) determining not to execute the update process when a state in which a detection value of a control output or a control input of the process substantially matches a first reference value continues for a reference time during the learning period;
The first reference value is a detected value of a control output or control input of the process, or an upper detection limit or lower detection limit value of a sensor that detects the control output or control input of the process, at a reference time within the learning period.
(2) If the variance or standard deviation of the detected value of the control output or control input of the process is less than a reference value during a reference period, which is at least a part of the learning period, it is decided not to execute the update process.
(3) determining not to execute the update process when a total time during which the detection value of the control output or the control input of the process substantially matches a second reference value during the learning period is equal to or longer than a reference time;
The second reference value is an upper or lower detection limit value of a sensor that detects a control output or a control input of the process. A model predictive control method for a process executed by an information processing device, comprising:
storing the process model;
performing model predictive control of a process using the process model;
accumulating detection values of control outputs and control inputs of the process as time series data;
executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
determining whether to perform the update process;
Including,
Performing the update process unless it is determined not to perform the update process;
a model prediction method comprising: determining not to execute the update process when a target value of a control output of the process changes within the learning period, if the amount of change in the target value of the control output of the process is less than a reference value, and if the detection value of the control output of the process substantially matches the target value over a period of the learning period excluding a tracking period after the change in the target value. A model predictive control method for a process executed by an information processing device, comprising:
storing the process model;
performing model predictive control of a process using the process model;
accumulating detection values of control outputs and control inputs of the process as time series data;
executing an update process for updating model parameters of the process model based on the time-series data accumulated during a learning period;
determining whether to perform the update process;
Including,
Performing the update process unless it is determined not to perform the update process;
a model prediction method in which, when a target value of a control output of the process changes within the learning period, a determination is made not to execute the update process if an evaluation value indicating the tracking of the control output of the process to the changed target value is within a predetermined reference range.

Images