Performance evaluating device, performance evaluating method and program

Abstract

To provide a technique that highly precisely evaluates the control performance of a controller.SOLUTION: A performance evaluating device that evaluates the control performance of a controller which controls an object to be controlled includes: an obtaining unit that obtains an actual operation amount representing an operation amount of the controller relative to the object to be controlled, and an actual control amount representing a controlled amount of the object to be controlled; a model parameter estimating unit that calculates, based on the actual operation amount and the actual control amount, an estimated value of a model parameter of the object to be controlled; a virtual control unit that calculates, when a virtual control amount representing a control amount in a virtual situation is input, a virtual operation amount representing an operation amount in a virtual situation in such a way that the virtual control amount follows an aiming value of the actual control amount by a predetermined control scheme; a plant simulating unit that calculates, when the virtual operation amount is input, the virtual control amount by simulating the object to be controlled based on the estimated value of the model parameter; and a performance evaluating unit that calculates an evaluation index value for comparing the control performance between the controller and the control scheme based on the aiming value, the virtual operation amount, the virtual control amount, the actual operation amount, and the actual control amount.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a performance evaluation apparatus, a performance evaluation method, and a program.

Background Art

[0002] Techniques for evaluating the control performance of a controller that controls a control target such as a plant have been conventionally known. For example, Patent Document 1 discloses a technique in which a control parameter of a PID controller is determined by performing a simulation using a process model obtained by identifying the response of a control target and a control model of the PID controller that controls the control target, and then the control performance of the PID controller with the set control parameter is monitored in real time.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the prior art, the accuracy in evaluating the control performance of a controller may decrease. For example, when the deviation between the control target and the model obtained by identifying its response is large, the accuracy in evaluating the control performance decreases.

[0005] The present disclosure has been made in view of the above points, and an object thereof is to provide a technique capable of evaluating the control performance of a controller with high accuracy.

Means for Solving the Problems

[0006] A performance evaluation apparatus according to an aspect of the present disclosure is a performance evaluation apparatus that evaluates the control performance of a controller that controls a control target, and includes an acquisition unit that acquires an actual operation amount representing an operation amount of the controller with respect to the control target and an actual control amount representing a control amount of the control target, a model parameter estimation unit that sequentially calculates an estimated value of model parameters of the control target based on the actual operation amount and the actual control amount, a virtual control unit that, when a virtual control amount representing a virtual control amount is input, calculates a virtual operation amount representing a virtual operation amount so that the virtual control amount follows a target value with respect to the actual control amount by a predetermined control method, a plant simulation unit that calculates the virtual control amount by simulating the control target based on the estimated value of the model parameters when the virtual operation amount is input, and a performance evaluation unit that calculates an evaluation index value for comparing the control performance between the controller and the control method based on the target value, the virtual operation amount, the virtual control amount, the actual operation amount, and the actual control amount.

Advantages of the Invention

[0007] A technique is provided that can accurately evaluate the control performance of a controller.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

Figure 22

Figure 23

Figure 24

Figure 25

Figure 26

Figure 27

Figure 28

Figure 29

Mode for Carrying Out the Invention

[0009] Hereinafter, the first to third embodiments of the present invention will be described in detail with reference to the drawings. Hereinafter, a plant is assumed as a control target, and a performance evaluation apparatus 10 that can highly accurately evaluate the control performance of a controller that controls the control target plant will be described.

[0010] Note that the performance evaluation apparatus 10 described below is assumed to be realized by a computing device such as a general-purpose server or a PC (personal computer) as an example. However, this is only an example, and the performance evaluation apparatus 10 may be realized by an edge device such as a PLC (Programmable Logic Controller) or a DCS (Distributed Control System).

[0011] Also, assuming a plant as a control target is an example, and the control target is not limited to a plant, and it is possible to assume any device, apparatus, facility, etc. that is a control target of the controller.

[0012] Here, in the prior art such as the above-mentioned Patent Document 1, the accuracy in evaluating control performance may decrease. For example, when the deviation between the controlled plant and the plant model obtained by identifying its response is large, the accuracy in evaluating control performance decreases. On the other hand, the controlled plant may have non-linearity, time-variation, etc., and it is very difficult to create a plant model that faithfully reproduces complex behaviors such as non-linearity and time-variation. Therefore, in the performance evaluation device 10 described below, the difficulty is solved by sequentially estimating the plant model from the measured values during the operation of the controlled plant.

[0013] Also, in the prior art such as the above-mentioned Patent Document 1, simulation is performed when determining control parameters, but the control performance cannot be compared between the simulation and the actual control. Therefore, in the performance evaluation device 10 described below, this is solved by directly comparing the manipulated variable and the controlled variable obtained when actually controlling the controlled plant with the manipulated variable and the controlled variable obtained during the simulation. Hereinafter, the manipulated variable and the controlled variable obtained when actually controlling the controlled plant will also be referred to as the "actual manipulated variable" and the "actual controlled variable", respectively. On the other hand, the manipulated variable and the controlled variable obtained during the simulation will also be referred to as the "virtual manipulated variable" and the "virtual controlled variable", respectively. Also, hereinafter, actually controlling the controlled plant will also be referred to as "actual control", and simulating (imitating) the control of the controlled plant using the plant model will also be referred to as "virtual control".

[0014] Also, generally, during plant operation, a situation may occur where the controlled variable and the manipulated variable fluctuate due to a change in the target value or an external disturbance in real time. However, in the simulation in the prior art such as the above-mentioned Patent Document 1, such a situation is not considered. For this reason, there may be a large deviation between the actual control and the virtual control. Therefore, in the performance evaluation device 10 described below, when a certain period has elapsed, or when the actual control and the virtual control deviate to a certain extent, etc., the actual manipulated variable and the actual controlled variable are set for the virtual manipulated variable and the virtual controlled variable, respectively, to prevent a situation where the actual control and the virtual control deviate greatly.

[0015] [First Embodiment] Hereinafter, the first embodiment will be described.

[0016] <Hardware Configuration Example of Performance Evaluation Device 10 According to the First Embodiment> A hardware configuration example of the performance evaluation device 10 according to the first embodiment will be described with reference to FIG. 1. FIG. 1 is a diagram showing an example of the hardware configuration of the performance evaluation device 10 according to the first embodiment.

[0017] As shown in FIG. 1, the performance evaluation device 10 according to the first embodiment includes an input device 11, a display device 12, an external I / F 13, a communication I / F 14, a RAM (Random Access Memory) 15, a ROM (Read Only Memory) 16, an auxiliary storage device 17, and a processor 18. These hardware components are communicably connected to each other via a bus 19.

[0018] The input device 11 is, for example, a keyboard, a mouse, a touch panel, a physical button, etc. The display device 12 is, for example, a display, a display panel, etc. Note that the performance evaluation device 10 may not have at least one of the input device 11 and the display device 12, for example.

[0019] The external I / F 13 is an interface with an external device such as a recording medium 13a. Examples of the recording medium 13a include a CD (Compact Disc), a DVD (Digital Versatile Disk), an SD memory card (Secure Digital memory card), a USB (Universal Serial Bus) memory card, etc.

[0020] The communication I / F 14 is an interface for the performance evaluation device 10 to communicate with other devices, other apparatuses, other terminals, etc. The RAM 15 is a volatile semiconductor memory (storage device) that temporarily holds programs and data. The ROM 16 is a non-volatile semiconductor memory (storage device) that can hold programs and data even when the power is turned off. The auxiliary storage device 17 is a non-volatile storage device such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), or a flash memory, in which programs and data are stored. The processor 18 is various arithmetic devices such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit).

[0021] By having the hardware configuration shown in FIG. 1, the performance evaluation device 10 according to the first embodiment can realize various processes to be described later. However, the hardware configuration shown in FIG. 1 is an example, and the hardware configuration of the performance evaluation device 10 according to the first embodiment is not limited thereto. For example, the performance evaluation device 10 according to the first embodiment may have a plurality of auxiliary storage devices 17 and a plurality of processors 18, may not have a part of the illustrated hardware, or may have various hardware other than the illustrated hardware.

[0022] <Functional configuration example of the performance evaluation device 10 according to the first embodiment> A functional configuration example of the performance evaluation device 10 according to the first embodiment will be described with reference to FIG. 2. FIG. 2 is a diagram showing an example of the functional configuration of the performance evaluation device 10 according to the first embodiment.

[0023] As shown in FIG. 2, the performance evaluation apparatus 10 according to the first embodiment includes a measurement unit 101, a model parameter estimation unit 102, a real / virtual switching unit 103, a virtual control unit 104, a plant simulation unit 105, a performance evaluation recording unit 106, a display control unit 107, and a timer 108. Each of these units is realized, for example, by processing executed by one or more programs installed in the performance evaluation apparatus 10 on a processor 18 or the like. Further, the performance evaluation apparatus 10 according to the first embodiment has a history DB 109. The history DB 109 is realized, for example, by a storage area such as an auxiliary storage device 17. Note that the history DB 109 may be realized by a storage area of a storage device communicably connected to the performance evaluation apparatus 10, for example.

[0024] The measurement unit 101 measures (samples) the actual operation amount u output from the control device 20 to the controlled plant 30 and the actual control amount y of the controlled plant 30 every sampling period T, and outputs the measured actual operation amount u and actual control amount y. Here, the control device 20 is a controller that controls the controlled plant 30 by a certain predetermined control method (e.g., PID control, model predictive control, etc.). The controlled plant 30 is various plants to be controlled (e.g., petrochemical plants, power generation plants, waste treatment plants, steel plants, food plants, etc.). Examples of the actual control amount y include temperature, pressure, flow rate, etc. Examples of the actual operation amount u include the opening / closing amount of a valve that controls the flow rate of a heat medium, etc. c Hereinafter, when explicitly indicating the index t of the time when the actual operation amount u is measured, it is expressed as u(t). Similarly, when explicitly indicating the index t of the time when the actual control amount y is measured, it is expressed as y(t). For other variables as well, when explicitly indicating the time index t, “(t)” shall be written after the variable. Note that the time index (hereinafter referred to as the time index) is updated as t←t + 1 every sampling period T, for example.

[0025] Hereinafter, when explicitly indicating the index t of the time when the actual operation amount u is measured, it is expressed as u(t). Similarly, when explicitly indicating the index t of the time when the actual control amount y is measured, it is expressed as y(t). For other variables as well, when explicitly indicating the time index t, “(t)” shall be written after the variable. Note that the time index (hereinafter referred to as the time index) is updated as t←t + 1 every sampling period T, for example. c Hereinafter, when explicitly indicating the index t of the time when the actual operation amount u is measured, it is expressed as u(t). Similarly, when explicitly indicating the index t of the time when the actual control amount y is measured, it is expressed as y(t). For other variables as well, when explicitly indicating the time index t, “(t)” shall be written after the variable. Note that the time index (hereinafter referred to as the time index) is updated as t←t + 1 every sampling period T, for example.

[0026] The model parameter estimation unit 102 samples at a sampling period T cEach time, using the actual manipulated variable u and the actual controlled variable y as inputs, an estimated value θ of the model parameters of the plant model est (hereinafter also referred to as the estimated value θ of the model parameters est ).) is calculated. Note that the plant model is a model representing the plant response and is expressed by a function or the like including the model parameters to be estimated. Hereinafter, the function representing the plant model is also referred to as the "plant response function". Examples of the plant model include an ARMA model, an ARMAX model, a neural network model, and the like.

[0027] The actual / virtual switching unit 103, at each sampling period T c and each switching period K s using the actual manipulated variable u, the actual controlled variable y, the virtual manipulated variable u s and the virtual controlled variable y s and the target value r as inputs, outputs a reference manipulated variable u b and a reference controlled variable y b and a switching flag f s . At this time, the actual / virtual switching unit 103, at each switching period K s or when the deviation between the actual control and the virtual control becomes large, sets the switching flag f s to ON and then sets the actual manipulated variable u and the actual controlled variable y to the reference manipulated variable u b and the reference controlled variable y b respectively. On the other hand, the actual / virtual switching unit 103, at other times (that is, when the switching flag f s is set to OFF), sets the virtual manipulated variable u b and the virtual controlled variable y b to the reference manipulated variable u s and the virtual controlled variable y s respectively. Here, the switching period K s is the period for switching the switching flag f s from OFF to ON, and its value is set in advance. Also, the reference manipulated variable u b is the manipulated variable used as a reference when calculating the virtual manipulated variable u s and the virtual controlled variable y s . Similarly, the reference controlled variable y b is the controlled variable used as a reference when calculating the virtual manipulated variable u sand the virtual control quantity y s is the control quantity used as a reference when calculating. When the switching flag is set to ON, the reference operation quantity u b and the reference control quantity y b are set to the actual operation quantity u and the actual control quantity y respectively. Therefore, the virtual operation quantity u s coincides with the actual operation quantity u, and the virtual control quantity y s coincides with the actual control quantity y. On the other hand, when the switching flag is set to OFF, the reference operation quantity u b and the reference control quantity y b are set to the virtual operation quantity u s and the virtual control quantity y s respectively. Therefore, the virtual operation quantity u s is an independent value from the actual operation quantity u, and the virtual control quantity y s is an independent value from the actual control quantity y.

[0028] The virtual control unit 104 calculates the virtual operation quantity u c every sampling period T s using the virtual control quantity y b , the reference operation quantity u b , the reference control quantity y s , the switching flag f s , the control parameter p, the target value r as inputs. At this time, the virtual control unit 104 functions as a simulator that simulates (emulates) a virtual controller, and calculates the virtual operation quantity u s so that the virtual control quantity y s follows the target value r.

[0029] The plant simulation unit 105 calculates the virtual control quantity y c every sampling period T s using the virtual operation quantity u b , the reference operation quantity u b , the reference control quantity y s , the switching flag f est , and the model parameter estimated value θ s as inputs. At this time, the plant simulation unit 105 functions as a simulator that simulates (emulates) the controlled plant 30, and calculates the virtual control quantity y s by simulating the controlled plant 30.

[0030] The performance evaluation recording unit 106 samples at a sampling period T c each time, and uses the target value r, the virtual operation amount u s and the virtual control amount y s as well as the actual operation amount u and the actual control amount y as inputs to calculate evaluation index values for evaluating the performance of actual control and virtual control, and stores these evaluation index values in the history DB 109.

[0031] The display control unit 107 samples at a sampling period T c each time, and causes a display device 12 such as a display to display a screen (hereinafter also referred to as a performance evaluation comparison screen) in which the evaluation index values stored in the history DB 109 are included in a comparable manner. However, causing the performance evaluation comparison screen to be displayed on the display device 12 is just an example, and the display control unit 107 may cause the performance evaluation comparison screen to be displayed on a display provided in another terminal or the like that is communicably connected to the performance evaluation device 10, for example.

[0032] The timer 108 samples at a sampling period T c each time, and is an operation trigger for operating the measurement unit 101, the model parameter estimation unit 102, the real / virtual switching unit 103, the virtual control unit 104, the plant simulation unit 105, the performance evaluation recording unit 106, and the display control unit 107. Note that the sampling period T c is the period for measuring (sampling) the actual operation amount u and the actual control amount y, and its value is set in advance.

[0033] The history DB 109 stores the history of the performance index values calculated by the performance evaluation recording unit 106.

[0034] <Model Parameter Estimation Process> Hereinafter, a process for estimating the model parameters of the plant model (that is, the model parameter estimation value θ estAn example of the process of calculating (the process) will be described with reference to FIG. 3. FIG. 3 is a flowchart showing an example of the model parameter estimation process. Hereinafter, when executing the model parameter estimation process for a certain index j at the current time index t, using j as the index used when executing the model parameter estimation process, the case will be described. Also, hereinafter, as an example, the case of calculating the model parameter estimation value θ est will be described.

[0035] Step S101: The model parameter estimation unit 102 determines whether to initialize the model parameter estimation value θ est (j) and the covariance matrix P(j). Here, as cases where it is determined to initialize, for example, when calculating the model parameter estimation value θ est for the first time (that is, at the first calculation of the model parameter estimation value θ est ), when an initialization instruction is given by the user or the like, etc. can be cited.

[0036] When it is determined to initialize the model parameter estimation value θ est (j) and the covariance matrix P(j) (YES in step S101), the model parameter estimation unit 102 proceeds to step S102. On the other hand, when it is not determined to initialize the model parameter estimation value θ est (j) and the covariance matrix P(j) (NO in step S101), the model parameter estimation unit 102 proceeds to step S103.

[0037] Step S102: The model parameter estimation unit 102 initializes θ est (0)=θ 0 and P(0)=I, and further initializes the previous value y of the actual control amount -1 ←y(t) and initializes j←0. Here, θ 0 is the model parameter estimation value θ already set in the plant response function estIt may use [a certain value], or may use the initial value preset by the user or the like, or may use a fixed initial value such as a vector with all elements being 0. Also, I may be an identity matrix or an arbitrary matrix determined in advance.

[0038] Step S103: The model parameter estimation unit 102 updates j←j + 1, and updates the state vector φ(j) based on the state vector φ(j - 1), the actual operation amount u(t), the actual control amount y(t), and the previous value y of the actual control amount. -1 For example, the model parameter estimation unit 102 updates the state vector φ(j) as follows.

[0039]

Equation

[0040] Note that, for example, considering the disturbance w for the past L points of the disturbance w to the controlled plant 30, the state vector φ(j) may be updated. In this case, in addition to y(j) and u(j), the state vector φ(j) is expanded to a vector having L disturbances w(j) as elements.

[0041] Step S104: The model parameter estimation unit 102 estimates the model parameter value θ estBased on (j - 1), state vector φ(j), and actual control amount y(t), the prediction error ε(j) is calculated. Note that the estimated model parameter value θ est (j - 1) is the estimated value of the model parameters estimated in the previous time (i.e., when j - 1).

[0042] The model parameter estimation unit 102 calculates the prediction error ε(j) as follows, for example.

[0043] y(j) = y(t) ε(j) = y(j) - φ(j) Τ θ est (j - 1) Here, Τ represents the symbol for transpose.

[0044] Step S105: The model parameter estimation unit 102 updates the covariance matrix P(j). The model parameter estimation unit 102 updates the covariance matrix P(j) as follows, for example.

[0045]

Equation

[0046] Step S106: The model parameter estimation unit 102 determines whether to update the estimated model parameter value θ est Here, cases where it is determined to update the estimated model parameter value θ est include, for example, cases where the estimated model parameter value θ est has not been updated until a predetermined period has elapsed since the initialization in step S102 above, cases where an update instruction is given by the user or the like, etc.

[0047] The estimated model parameter value θ estIf it is determined to update (YES in step S106), the model parameter estimation unit 102 proceeds to step S107. On the other hand, if it is not determined to update the model parameter estimated value θ est (NO in step S106), the model parameter estimation unit 102 proceeds to step S108.

[0048] Step S107: The model parameter estimation unit 102 updates the model parameter estimated value θ est (j). The model parameter estimation unit 102 updates the model parameter estimated value θ est (j), for example, as follows.

[0049] [Number] [Actual / virtual switching process] Hereinafter, an example of a process of determining whether to switch the switching flag f s and, according to the determination result, setting the virtual operation amount u s and the virtual control amount y s , or the actual operation amount u and the actual control amount y to the reference operation amount u b and the reference control amount y b will be described with reference to FIG. 4. FIG. 4 is a flowchart showing an example of the actual / virtual switching process. Hereinafter, regarding the index k 1 used when executing the actual / virtual switching process, a case where the actual / virtual switching process is executed for a certain index k 1 with the current time index t will be described.

[0050] Step S201: The actual / virtual switching unit 103 determines whether the switching flag f s (k 1 ) is ON.

[0051] If it is determined that the switching flag f s (k 1 ) is ON (YES in step S201), the actual / virtual switching unit 103 proceeds to step S202. On the other hand, if the switching flag f s (k 1) If it is not determined that (is ON) (NO in step S201), the actual / virtual switching unit 103 proceeds to step S203.

[0052] Step S202: The actual / virtual switching unit 103 sets the determination state E s (0) = 0 and E r (0) to 0 and further initializes the switching flag f s (0) to OFF and initializes k 1 ← 0.

[0053] Step S203: The actual / virtual switching unit 103 updates k 1 ← k 1 +1 and updates the determination states E s (k 1 ) and E r (k 1 ) as follows.

[0054]

Number

[0055] Step S204: The actual / virtual switching unit 103 determines whether any of the following conditions (1) or (2) is satisfied. If any of the following conditions (1) or (2) is satisfied, f s (k 1 ) is set to ON. Note that if neither of the following conditions (1) and (2) is satisfied, f s (k 1 ) is not updated (in this case, f s (k 1 ) remains OFF.).

[0056] (1) k 1 ≥ K s is true.

[0057] (2) E s (k 1 ) is a predetermined multiple or more of E r (k 1 ), that is, it is as follows.

[0058]

Number

[0059] Note that the condition shown in (1) above means that the index k 1 is greater than or equal to the switching period K s , and the switching flag f s is set to ON at a fixed period. On the other hand, the condition shown in (2) above means that the deviation between the actual control and the virtual control has become large, and when the deviation between the actual control and the virtual control becomes large, the switching flag f s is set to ON. Note that the case where the condition shown in (2) above is satisfied includes, for example, the case where the simulation (simulation) by the plant simulation unit 105 and the behavior of the control target plant 30 are extremely different. However, in addition to the cases where the conditions shown in (1) and (2) above are satisfied, for example, f s (k 1 ) = ON may be set according to an instruction from a user or the like.

[0060] Step S205: The actual / virtual switching unit 103 determines whether the switching flag f s is ON.

[0061] If it is determined that the switching flag f s (k 1 ) is ON (YES in step S205), the actual / virtual switching unit 103 proceeds to step S206. On the other hand, if it is not determined that the switching flag f s (k 1 ) is ON (NO in step S205), the actual / virtual switching unit 103 proceeds to step S207.

[0062] Step S206: The actual-virtual switching unit 103 sets u b (t) ← u(t), y b (t) ← y(t), f s (t) ← f s (k 1 ). That is, the actual-virtual switching unit 103 sets the actual operation amount u(t) and the actual control amount y(t) for the reference operation amount u b (t) and the reference control amount y b (t), respectively, and sets the switching flag f s (t) for the switching flag f s (k 1 ).

[0063] Step S207: The actual-virtual switching unit 103 sets u b (t) ← u s (t), y b (t) ← y s (t), f s (t) ← f s (k 1 ). That is, the actual-virtual switching unit 103 sets the virtual operation amount u b (t) calculated by the virtual control unit 104 for the reference operation amount u s (t), and sets the virtual control amount y b (t) calculated by the plant simulation unit 105 for the reference control amount y s (t). Also, the actual-virtual switching unit 103 sets the switching flag f s (t) for the switching flag f s (k 1 ).

[0064] Note that u b (t) and y b (t) set for u s (t) and y s (t) by the actual-virtual switching unit 103 are the values where the latest value at the operation time point of the actual-virtual switching unit 103 is held by zero-order hold. Therefore, in the state where it has not been updated by the virtual control unit 104, u s (t) = u s (k 2The index k is the previous value, such as (k - 1). 2 When it becomes the previous value and is not updated by the plant simulation unit 105, y s (t) = y s (k 3 The index k is the previous value, such as (k - 1). 3 Note that it becomes the previous value. For example, u s (t) and y s (t) are values written to the specified address of the shared memory, and the actual / virtual switching unit 103 can be realized with a configuration in which the value of the specified address of the shared memory at that time t is read.

[0065] <Virtual control process> Hereinafter, an example of the process for calculating the virtual operation amount u s will be described with reference to FIG. 5. FIG. 5 is a flowchart showing an example of the virtual control process. Hereinafter, regarding the index used when executing the virtual control process as k 2 and the current time index t, the case of executing the virtual control process for a certain index k 2 will be described. In the following, as an example, the case where the virtual control unit 104 simulates a virtual controller by PID control will be described.

[0066] Step S301: The virtual control unit 104 determines whether the switching flag f s (t) is ON.

[0067] If it is determined that the switching flag f s (t) is ON (YES in step S301), the virtual control unit 104 proceeds to step S302. On the other hand, if it is not determined that the switching flag f s (t) is ON (NO in step S301), the virtual control unit 104 proceeds to step S303.

[0068] Step S302: The virtual control unit 104 calculates the virtual target deviation e s (0) = r(t) - y bInitialize (t) and initialize the control state η(0) as follows. Further, set k 2 ← 0.

[0069]

Number

[0070] Step S303: The virtual control unit 104 updates k 2 ← k 2 + 1, calculates e s (k 2 ) = r(t) - y b (t), and updates the control state η(k 2 ) as follows.

[0071]

Number

[0072] Step S304: The virtual control unit 104 calculates the virtual operation amount u s (t) as follows.

[0073] u s (k 2 ) = η(k 2 ) Τ p u s (t) ← u s (k 2 ) Here, p is a control parameter and is represented as follows.

[0074] [Number] That is, the control parameter p is a vector composed of three elements for simulating PID control.

[0075] In the above virtual control process, the case where the virtual control unit 104 simulates a virtual controller by PID control has been described, but this is just an example. The virtual control unit 104 may simulate a virtual controller by, for example, other control methods. For example, taking the function or algorithm for calculating the operation change amount by the method described in Japanese Patent Application Laid-Open No. 2020-21411 as f(·;p), the operation change amount du s (k 2 ) = f(y b (k 2 ), u b (k 2 ); p) is calculated, and then u s (k 2 ) = u b (k 2 - 1) + du s (k 2 ), and u s (t) ← u s (k 2 ) is used to calculate the virtual operation amount u s (t). When this method is used, when the switching flag f s (t) is ON, the prediction time series memory unit described in Japanese Patent Application Laid-Open No. 2020-21411 is initialized. However, this is just an example, and f(·;p) may be a function or algorithm for calculating the operation change amount by a method other than the method described in Japanese Patent Application Laid-Open No. 2020-21411.

[0076] <Plant Simulation Process> Hereinafter, an example of the process for calculating the virtual control amount y s will be described with reference to FIG. 6. FIG. 6 is a flowchart showing an example of the plant simulation process. Hereinafter, using the index k 3 used when executing the plant simulation process, at the current time index t, a certain index k 3A case of executing plant simulation processing will be described.

[0077] Step S401: The plant simulation unit 105 determines whether the switching flag f s (t) is ON.

[0078] The switching flag f s (t) is determined to be ON (YES in step S401), the plant simulation unit 105 proceeds to step S402. On the other hand, if the switching flag f s (t) is not determined to be ON (NO in step S401), the plant simulation unit 105 proceeds to step S403.

[0079] Step S402: The plant simulation unit 105 initializes u(0) = u b (t), y(0) = y b (t), u s (0) = u b (t), y s (0) = y b (t) and initializes the state vector ψ(0) as follows, and further initializes k 3 ←0.

[0080] [Equation] Step S403: The plant simulation unit 105 updates k 3 ←k 3 +1 and updates the state vector ψ(k 3 ) as follows.

[0081] [Equation] Here, y(k 3 -1) is set to y s (k 3 -1), and u(k 3 ) is set to u b (t). Also, y(k 3 -2), ···, y(k 3 -N) and u(k 3-1), ···, u(k 3 -M) is set to the same value as the state vector ψ(k 3 -1) (that is, the y(k 3 contained in the state vector ψ(k 3 -2), ···, y(k 3 -N) and u(k 3 -1), ···, u(k 3 -M) are each set.).

[0082] Step S404: The plant simulation unit 105 calculates the virtual control amount y s (t) as follows.

[0083] y s (k 3 ) = ψ(k 3 ) Τ θ est (t) y s (t) ← y s (k 3 ) In this embodiment, it is assumed that the plant simulation process is executed after the virtual control process is executed at each current index t. However, this is just an example, and for example, the virtual control process may be executed after the plant simulation process is executed at each time index t.

[0084] <Performance Evaluation and Recording Process> Hereinafter, an example of a process of calculating evaluation index values and storing those evaluation index values in the history DB 109 will be described with reference to FIG. 7. FIG. 7 is a flowchart showing an example of the performance evaluation and recording process. Hereinafter, when the performance evaluation and recording process is executed for a certain index k 4 at the current time index t, the case will be described. Also, hereinafter, as an example, the average absolute error v 4 between the virtual control amount y s and the target value, the average absolute error v 1s between the actual control amount y and the target value r, and the virtual control amount y 1r and the virtual control amount y sand the root mean square error (RMSE) v of the target value r 2s , the root mean square error v of the actual control amount y and the target value r 2r , the virtual operation amount u s of the average operation change amount v 3s , the average operation change amount v of the actual operation amount u 3r The case of calculating six evaluation index values will be described. In the following, v 1s and v 1r are referred to as "target deviation MAE", v 2s and v 2r are referred to as "target deviation RMSE", v 3s and v 3r will also be referred to as "average operation change amount".

[0085] Step S501: The performance evaluation recording unit 106 determines whether the switching flag f s (t) is ON.

[0086] If it is determined that the switching flag f s (t) is ON (YES in step S501), the performance evaluation recording unit 106 proceeds to step S502. On the other hand, if it is not determined that the switching flag f s (t) is ON (NO in step S501), the performance evaluation recording unit 106 proceeds to step S503.

[0087] Step S502: The performance evaluation recording unit 106 initializes v 1s (0)=0, v 1r (0)=0, v 2s (0)=0, v 2r (0)=0, v 3s (0)=0, v 3r (0)=0 and initializes k 4 ←0.

[0088] Step S503: The performance evaluation recording unit 106 updates k 4 ←k 4 +1 and updates r(k 4 )←r(t), y(k 4 )←y(t), u(k 4) ← u(t), y s (k 4 ) ← y s (t), u s (k 4 ) ← u s (t) and update, and each evaluation index value v 1s (k 4 ), v 1r (k 4 ), v 2s (k 4 ), v 2r (k 4 ), v 3s (k 4 ), v 3r (k 4 ) is updated.

[0089]

Number

[0090]

Number

[0091]

Number

[0092] <Performance Evaluation Comparison Screen> Hereinafter, a specific example of the performance evaluation comparison screen displayed on the display device 12 or the like by the display control unit 107 will be described.

[0093] · Specific Example 1 A specific example 1 of the performance evaluation comparison screen is shown in FIG. 8. In the performance evaluation comparison screen 1100 shown in FIG. 8, the target deviation MAEv 4 at the current index k 1s (k 4 ) and v 1r (k 4 ), the target deviation RMSEv 2s (k 4 ) and v 2r (k 4 ), and the average operation change amount v 3s (k 4 ) and v 3r (k 4 ) are displayed so as to be comparable.

[0094] In addition, in the performance evaluation comparison screen 1100 shown in FIG. 8, the improvement rates a 1 , a 2 , a 3 of the virtual control with respect to the actual control are displayed. Here, a 1 is the improvement rate of the target deviation MAE, and is calculated by ((v 1r (k 4 ) - v 1s (k 4 )) / v 1r (k 4 ) × 100 [%]. a 2 is the improvement rate of the target deviation RMSE, and is calculated by ((v 2r (k 4 ) - v 2s (k 4 )) / v 2r (k 4 ) × 100 [%]. a 3 is the improvement rate of the average operation change amount, and is calculated by ((v 3r (k 4 ) - v 3s (k 4 )) / v 3r (k 4 ) × 100 [%]. Note that in FIG. 8, "▲" indicates that the improvement rate is negative.

[0095] Thereby, for example, it becomes possible to know how much improvement in control performance can be expected when the actual control is replaced with the virtual control.

[0096] · Specific Example 2 A specific example 2 of the performance evaluation comparison screen is shown in Fig. 9. In the performance evaluation comparison screen 1200 shown in Fig. 9, for each evaluation period, the improvement rate of the target deviation MAE, the improvement rate of the target deviation RMSE, and the improvement rate of the average operation change amount in that evaluation period are displayed. Here, the evaluation period means the period after the switching flag f s becomes ON, and then immediately after that, the switching flag f s becomes OFF and then becomes ON again. In other words, after k 4 =0, for a certain k 4 =k 4 ' when k 4 is initialized to 0, the period from k 4 =0 to k 4 ' is the evaluation period. Note that at the start of a certain time index t, k 1 =k 2 =k 3 =k 4 should be noted.

[0097] In the performance evaluation comparison screen 1200 shown in Fig. 9, as an example, the improvement rates a 11 ~a 31 of the standard deviation MAE, the improvement rates a 12 ~a 32 of the target deviation RMSE, and the improvement rates a 13 ~a 33 of the average operation change amount in three evaluation periods are displayed.

[0098] Thereby, for example, it becomes possible to know how each improvement rate changes for each evaluation period.

[0099] · Specific Example 3 A specific example 3 of the performance evaluation comparison screen is shown in Fig. 10. In the performance evaluation comparison screen 1300 shown in Fig. 10, for each evaluation period, the target deviation MAE, the target deviation RMSE, and the average operation change amount between the actual control and the virtual control in that evaluation period are displayed as a graph in time series.

[0100] On the performance evaluation comparison screen 1300 shown in FIG. 10, as an example, the time series of the target deviation MAE, target deviation RMSE, and average of the amount of operation change of the actual control and the virtual control in three evaluation periods are displayed.

[0101] As a result, since the evaluation index values of the actual control and the virtual control can be compared in time series, it becomes possible to visually grasp the effect of the virtual control. In addition, on the performance evaluation comparison screen 1300 shown in FIG. 10, since a dividing line indicating the division of the evaluation period is also displayed, it is also possible to easily grasp how the evaluation index value is reset to 0 for each evaluation period.

[0102] Note that on the performance evaluation comparison screen, not only the evaluation index value but also, for example, the target value r, the actual operation amount u, the virtual operation amount u s 、the actual control amount y, the virtual control amount y s 、the model parameter estimated value θ est etc. may be displayed as a graph in time series in whole or in part.

[0103] [Second Embodiment] Hereinafter, the second embodiment will be described. In the second embodiment, mainly, the differences from the first embodiment will be described, and the description of the same components as those in the first embodiment will be omitted.

[0104] [Functional Configuration Example of the Performance Evaluation Apparatus 10 According to the Second Embodiment] A functional configuration example of the performance evaluation apparatus 10 according to the second embodiment will be described with reference to FIG. 11. FIG. 11 is a diagram showing an example of the functional configuration of the performance evaluation apparatus 10 according to the second embodiment.

[0105] As shown in FIG. 11, the virtual control unit 104 of the performance evaluation apparatus 10 according to the second embodiment further inputs the model parameter estimated value θ est . As a result, the virtual control unit 104 can calculate the virtual control amount y est using, for example, the model parameter estimated value θ s by model predictive control or the like.

[0106] Also, the performance evaluation device 10 according to the second embodiment outputs the virtual operation amount u c to the switch 40 every sampling period T s (t). Here, in the switch 40, according to the setting information indicating which of the operation amount or the virtual operation amount u s output from the control device 20 is to be output to the controlled plant 30, either the operation amount or the virtual operation amount u s output from the control device 20 is output to the controlled plant 30 as the actual operation amount u. Note that the performance evaluation device 10 according to the second embodiment may have an "output unit" as a functional unit that outputs the virtual operation amount u s (t) to the switch 40.

[0107] Thus, for example, if it is determined from the result of referring to the performance evaluation comparison screen that good control can be achieved by virtual control, the switch 40 can output the virtual operation amount u s as the actual operation amount u to the controlled plant 30.

[0108] [Third Embodiment] Hereinafter, the third embodiment will be described. In the third embodiment, mainly, the differences from the first embodiment will be described, and the description of the same components as those in the first embodiment will be omitted.

[0109] <Functional Configuration Example of the Performance Evaluation Device 10 According to the Third Embodiment> A functional configuration example of the performance evaluation device 10 according to the third embodiment will be described with reference to FIG. 12. FIG. 12 is a diagram showing an example of the functional configuration of the performance evaluation device 10 according to the third embodiment.

[0110] As shown in FIG. 12, the performance evaluation device 10 according to the third embodiment has a data acquisition unit 110 instead of the measurement unit 101. The data acquisition unit 110 acquires the target value r(t), the actual operation amount u(t), and the actual control amount y(t) as performance data from the operation result DB501 provided in the database server 50.

[0111] Accordingly, for example, even when the plant 30 to be controlled has stopped or is being repaired, it is possible to compare the control performance of the actual control and the virtual control using past performance data.

[0112] [Embodiment] Hereinafter, Examples 1 to 5 of the performance evaluation apparatus 10 according to the first to third embodiments will be described.

[0113] The plant model of the plant 30 to be controlled is assumed to be represented by the following second-order ARMA model.

[0114] y(k)=a 1 y(k-1)+a 2 y(k-2)+b 0 u(k)+b 1 u(k-1)+b 2 u(k-2) Therefore, the state vector ψ(k 3 ) used in the plant simulation unit 105 is as follows.

[0115] [Equation] Also, the estimated value θ est (j) is as follows.

[0116] [Equation] Also, assume that the step response of the plant 30 to be controlled is as shown in FIG. 13.

[0117] When the control device 20 performs control by PID control, the manipulated variable u is calculated as follows in the continuous representation.

[0118] [Equation] On the other hand, in the discrete representation, the manipulated variable u is calculated as follows.

[0119]

Mathematics

[0120]

Mathematics

[0121] At this time, the control response when controlled by the control device 20 is shown in FIG. 14. As shown in FIG. 14, although the control amount y follows the target value r to a certain extent, the response is oscillatory and the tracking performance is not very good. Note that the disturbance w is constant and has no influence on the control performance.

[0122] ·Example 1 In the performance evaluation device 10 in Example 1, a sufficiently large value is set for the switching period K s , and it is assumed that the virtual control unit 104 uses the same PID control as the control device 20. However, only the integral gain k I is set to k I = 0.01, which is twice the value of the control device 20. Therefore, it can be said that in Example 1, the control response when the integral gain is doubled is being evaluated.

[0123] The target value r, virtual control amount y s , virtual operation amount u s , actual control amount u, and actual operation amount y in Example 1 are shown in FIG. 15. Also, the model parameter estimated value θ est in Example 1 is shown in FIG. 16. Also, the evaluation index value and index k 4 in Example 1 are shown in FIG. 17. Each time-series graph shown in FIGS. 15 to 17 is displayed on the display device 12 or the like by, for example, the display control unit 107. Note that in Example 1, since the switching period K s is sufficiently large, the switching flag f sIt is ON only at time t = 0 and then OFF.

[0124] As shown in FIG. 17, the target deviation MAEv 1r and v 1s and the target deviation RMSEv 2r and v 2s maintain lower values for virtual control than for actual control. On the other hand, the average operation change amount v 3r and v 3s has slightly higher values for virtual control than for actual control.

[0125] Also, as shown in FIG. 15, virtual control has better tracking performance for the target value r than actual control, and the target deviation MAEv 1r and v 1s and the target deviation RMSEv 2r and v 2s are consistent with the fact that virtual control is lower than actual control.

[0126] ·Example 2 In the performance evaluation device 10 in Example 2, 500 is set for the switching period K s , and it is assumed that the virtual control unit 104 uses the same PID control as the control device 20. However, only for the integral gain k I , it is assumed that a value k I = 0.01 which is twice that of the control device 20 is set.

[0127] The target value r, virtual control amount y s , virtual operation amount u s , actual control amount u, and actual operation amount y in Example 2 are shown in FIG. 18. Also, the model parameter estimated value θ est in Example 2 is shown in FIG. 19. Also, the evaluation index values and index k 4 in Example 2 are shown in FIG. 20. Each time-series graph shown in FIGS. 18 to 20 is displayed on a display device 12 or the like by, for example, the display control unit 107.

[0128] As shown in FIG. 20, for each switching period K s = 500, the index k 4It can be seen that it is reset to 0. Also, for each evaluation index value, the switching period K s is reset to 0 every 500.

[0129] Comparing with Example 1, in Example 2, the performance evaluations of the actual control and the virtual control are reversed. That is, in Example 2, the target deviations MAEv 1r and v 1s and the target deviation RMSEv 2r and v 2s maintain lower values for the actual control than for the virtual control, and the average of the operation change amounts v 3r and v 3s is slightly higher for the actual control than for the virtual control. This is because in Example 2, since the switching by the actual-virtual switching unit 103 is performed, recalculation based on the actual control amount y and the actual operation amount u is performed, and it is considered that the response closer to the actual control is obtained with the virtual control than in Example 1.

[0130] · Example 3 In the performance evaluation device 10 in Example 3, a sufficiently large value is set for the switching period K s , and it is assumed that the virtual control unit 104 performs model predictive control. Note that the control device 20 performs PID control.

[0131] The target value r, the virtual control amount y s , the virtual operation amount u s , the actual control amount u, and the actual operation amount y in Example 3 are shown in FIG. 21. Also, the model parameter estimated value θ est in Example 3 is shown in FIG. 22. Also, the evaluation index values and the index k 4 in Example 3 are shown in FIG. 23. The graphs of each time series shown in these FIGS. 21 to 23 are displayed on a display device 12 or the like by, for example, the display control unit 107.

[0132] As shown in FIG. 23, it can be seen that in Example 3, the target deviation MAE and the target deviation RMSE are significantly improved by the model predictive control.

[0133] · Example 4 In the performance evaluation apparatus 10 in Example 4, the switching period K s is set to 500, and it is assumed that the virtual control unit 104 performs model predictive control. Note that the control device 20 performs PID control.

[0134] The target value r, virtual control amount y s , virtual operation amount u s , actual control amount u, and actual operation amount y in Example 4 are shown in FIG. 24. Also, the model parameter estimated value θ est in Example 4 is shown in FIG. 25. Further, the evaluation index value and index k 4 in Example 4 are shown in FIG. 26. Each time-series graph shown in FIGS. 24 to 26 is displayed on the display device 12 or the like by the display control unit 107, for example.

[0135] As shown in FIGS. 24 and 26, since switching is performed by the actual-virtual switching unit 103, although the improvement amount is reduced compared to Example 3, it can be seen that the model predictive control has a target tracking performance superior to that of the PID control.

[0136] ·Example 5 In the performance evaluation apparatus 10 in Example 4, the switching period K s is set to 500, and it is assumed that both the virtual control unit 104 and the control device 20 perform model predictive control.

[0137] The target value r, virtual control amount y s , virtual operation amount u s , actual control amount u, and actual operation amount y in Example 5 are shown in FIG. 27. Also, the model parameter estimated value θ est in Example 5 is shown in FIG. 28. Further, the evaluation index value and index k 4 in Example 5 are shown in FIG. 29. Each time-series graph shown in FIGS. 27 to 29 is displayed on the display device 12 or the like by the display control unit 107, for example.

[0138] As shown in FIGS. 27 to 29, the model parameter estimated value θ estAfter the time t = 500 when it stabilizes, the evaluation index values of the actual control and the virtual control are almost equal, and the time series graphs of the actual control amount y and the virtual control amount y s also have a fairly similar shape.

[0139] Also, focusing on the intervals of time t = 2000 to 2500 and time t = 3500 to 4000, it can be confirmed that the target deviation that occurred when PID control was used as the control method of the control device 20 has been significantly improved by using model predictive control, and the target deviation MAE and target deviation RMSE in this interval have been improved.

[0140] [Summary] As described above, by using the performance evaluation device 10 according to each of the above embodiments, it is possible to perform highly accurate performance evaluation while estimating the model parameters of the plant model of the control target plant 30 online. In addition, it is possible to confirm the performance improvement by virtual control with respect to the actual control during the online operation of the control target plant 30 without changing the currently operating control device 20. Moreover, at this time, the performance of the actual control and the virtual control can be directly, quantitatively, and visually confirmed. In addition, for example, even when the target value is changed in real time and the actual operation amount u and the actual control amount y fluctuate accordingly, the switching operation by the actual-virtual switching unit 103 enables comparison of control performance closer to the actual operation state of the control target plant 30.

[0141] The present invention is not limited to the specifically disclosed above embodiments, and various modifications, changes, combinations with known technologies, etc. are possible without departing from the description of the claims.

Explanation of Reference Numerals

[0142] 10 Performance evaluation device 11 Input device 12 Display device 13 External I / F 13a Recording medium 14 Communication I / F 15 RAM 16 ROM 17 Auxiliary storage device 18 Processor 19 Bus 20 Control device 30 Plant to be controlled 40 Switch 50 Database server 101 Measurement unit 102 Model parameter estimation unit 103 Real / virtual switching unit 104 Virtual control unit 105 Plant simulation unit 106 Performance evaluation recording unit 107 Display control unit 108 Timer 109 History DB 110 Data acquisition unit 501 Operation result DB

Claims

1. A performance evaluation device for evaluating the control performance of a controller that controls a control target, an acquisition unit that acquires an actual operation amount representing an operation amount of the controller with respect to the control target and an actual control amount representing a control amount of the control target; a model parameter estimation unit that sequentially calculates an estimated value of model parameters of the control target based on the actual operation amount and the actual control amount; a virtual control unit that, when a virtual control amount representing a virtual control amount is input, calculates a virtual operation amount representing a virtual operation amount so that the virtual control amount follows a target value with respect to the actual control amount by a predetermined control method; a plant simulation unit that calculates the virtual control amount by simulating the control target based on the estimated value of the model parameters when the virtual operation amount is input; a performance evaluation unit that calculates an evaluation index value for comparing the control performance between the controller and the control method based on the target value, the virtual operation amount, the virtual control amount, the actual operation amount, and the actual control amount; A performance evaluation device having the above.

2. It has an actual-virtual switching unit that calculates a reference operation amount and a reference control amount respectively representing an operation amount and a control amount that are standards when calculating the virtual operation amount and the virtual control amount, The actual-virtual switching unit, when a predetermined condition is satisfied, sets the actual operation amount and the actual control amount with respect to the reference operation amount and the reference control amount respectively, and when the condition is not satisfied, sets the virtual operation amount and the virtual control amount with respect to the reference operation amount and the reference control amount respectively, The virtual control unit, further inputs the reference operation amount and the reference control amount, and calculates the virtual operation amount, The plant simulation unit, further inputs the reference operation amount and the reference control amount, and calculates the virtual control amount. The performance evaluation device according to claim 1.

3. The conditions include, when it becomes a predetermined switching cycle determined in advance, when the ratio of the average value of the deviation between the virtual control amount and the target value and the average value of the deviation between the actual control amount and the target value exceeds a predetermined ratio, when a predetermined instruction is given from the user, The performance evaluation device according to claim 2, including at least one of the above.

4. The performance evaluation unit, when the condition is satisfied, initializes the evaluation index value to 0. The performance evaluation device according to claim 3.

5. The performance evaluation unit, A performance evaluation device according to any one of claims 1 to 4, which calculates, as the evaluation index values, a first evaluation index value representing the mean absolute error between the target value and the virtual control amount, a second evaluation index value representing the mean absolute error between the target value and the actual control amount, a third evaluation index value representing the root mean square error between the target value and the virtual control amount, a fourth evaluation index value representing the root mean square error between the target value and the actual control amount, a fifth evaluation index value representing the average operation change amount of the virtual operation amount, and a sixth evaluation index value representing the average operation change amount of the actual operation amount.

6. The performance evaluation device according to claim 5, further comprising a display control unit that causes a display device to display the first evaluation index value and the second evaluation index value, the third evaluation index value and the fourth evaluation index value, and the fifth evaluation index value and the sixth evaluation index value in a comparable manner.

7. The performance evaluation device according to claim 1, wherein the control method includes a control method based on given control parameters and model predictive control based on estimated values of the model parameters.

8. The model parameters are parameters included in an ARMA model or an ARMAX model that models the response of the control object, The model parameter estimation unit calculates an estimated value of the model parameters by the recursive least squares method based on the actual operation amount and the actual control amount. The performance evaluation device according to claim 1.

9. The performance evaluation device according to claim 1, further comprising an output unit that outputs the virtual operation amount to a switch that outputs either the virtual operation amount or the actual operation amount to the control object according to a predetermined setting.

10. A computer that evaluates the control performance of a controller that controls a control object includes an acquisition procedure for acquiring an actual operation amount representing an operation amount of the controller for the control object and an actual control amount representing a control amount of the control object; a model parameter estimation procedure for sequentially calculating estimated values of the model parameters of the control object based on the actual operation amount and the actual control amount; a virtual control procedure for calculating a virtual operation amount representing a virtual operation amount so that the virtual control amount follows a target value for the actual control amount by a predetermined control method when a virtual control amount representing a virtual control amount is input; a plant simulation procedure for calculating the virtual control amount by simulating the control object based on the estimated value of the model parameters when the virtual operation amount is input. A performance evaluation procedure for calculating an evaluation index value for comparing control performance between the controller and the control method based on the target value, the virtual operation amount, the virtual control amount, the actual operation amount, and the actual control amount. A performance evaluation method for executing the above. **Claim 11** On a computer for evaluating the control performance of a controller that controls a control target, An acquisition procedure for acquiring an actual operation amount representing the operation amount of the controller for the control target and an actual control amount representing the control amount of the control target; A model parameter estimation procedure for sequentially calculating an estimated value of the model parameters of the control target based on the actual operation amount and the actual control amount; When a virtual control amount representing a virtual control amount is input, a virtual control procedure for calculating a virtual operation amount representing a virtual operation amount so that the virtual control amount follows a target value for the actual control amount by a predetermined control method; When the virtual operation amount is input, a plant simulation procedure for calculating the virtual control amount by simulating the control target based on the estimated value of the model parameters; A performance evaluation procedure for calculating an evaluation index value for comparing control performance between the controller and the control method based on the target value, the virtual operation amount, the virtual control amount, the actual operation amount, and the actual control amount; A program for causing the above to be executed.

Images