Plant control system and plant control method

Abstract

To provide a control system that is able to output an appropriate operation variable for quickly converging a controlled variable to a target value even when a prediction value of a model becomes inappropriate due to the influence of disturbance or delay.SOLUTION: A plant control system for calculating an optimum operation variable for causing a value of a controlled variable of a plant, as a controlled object, to follow a control target value thereof, includes: an optimal transition model storage unit that, based on a value of the controlled variable obtained from a sensor installed in the plant, stores an optimal transition model for calculating, based on a state transition probability model, an optimal transition value that is a value of a controlled variable in a next state of the plant; and operation variable calculation means that calculates an optimum operation variable by using a predetermined reference operation variable and the optimum transition value calculated by the optimum transition model.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a plant control system and a plant control method. [Background technology]

[0002] In recent years, AI technology that can realize control operations based on predictions has been actively developed in the plant field. Specifically, it is desirable to use AI technology to build a model that predicts the future state of the plant from existing operating data, and to be able to implement optimal control operations based on that model. The optimal control operation here refers to a control operation that quickly stabilizes the signal value of the plant, in other words, a control operation that causes the controlled signal, which indicates the state of the plant, to converge to the target value. Since AI technology can perform control operations based on predictions of the plant state, it is expected to achieve higher accuracy than conventional control such as PID control.

[0003] Conventionally, this type of technology is described in Patent Document 1. The method disclosed in Patent Document 1 makes it possible to calculate the plant state infinitely ahead and the future state of its surrounding environment in the form of a probability density distribution. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-159876 A Summary of the Invention [Problem to be solved by the invention]

[0005] While AI technology can predict plant conditions with high accuracy, it is not necessarily easy to calculate optimal control operation amounts based on the predicted results. In addition, since AI outputs a specific point among probabilistic behaviors as a predicted value, there is inevitably a discrepancy between the prediction and the actual measurement, and unless some kind of action is taken, the control performance may deteriorate.

[0006] In other words, when plant control using AI technology is being implemented, if an unexpected disturbance occurs, if there is a delay in the state transition, or if the plant state changes more than expected due to other factors, the predicted values ​​of the model built by the AI ​​may become inappropriate. This is because the past operating data used to build the predictive model does not include all changes in the plant state. In other words, if a rare change occurs in the plant state due to a delay or disturbance, it becomes difficult for the AI ​​to predict the future plant state.

[0007] Given this background, there is a need for a method that can calculate appropriate control inputs even when the predicted values ​​of the model constructed by AI become inappropriate due to disturbances or delays.

[0008] In other words, an object of the present invention is to provide a control system that can calculate an appropriate control operation amount that quickly converges a controlled amount to a target value even when the model prediction value becomes inappropriate due to the influence of disturbances or delays. [Means for solving the problem]

[0009] In order to solve the above problems, the plant control system of the present invention is a plant control system that calculates an optimal manipulated variable for making the value of a controlled variable of a plant that is a controlled object follow its control target value, and is configured to include an optimal transition model storage unit in which an optimal transition model is stored that calculates an optimal transition value, which is the value of the controlled variable of the next state of the plant, based on a state transition probability model based on the value of the controlled variable obtained from a sensor installed in the plant, and a manipulated variable calculation means that calculates the optimal manipulated variable using a predetermined reference manipulated variable and the optimal transition value calculated by the optimal transition model. Effect of the Invention

[0010] It is possible to provide a control system that can output an appropriate manipulated variable that quickly converges a controlled variable to a target value even when the model prediction value becomes inappropriate due to the influence of disturbances or delays. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram showing an example of the configuration of a plant control system according to an embodiment of the present invention. [Diagram 2] FIG. 4 is a diagram showing an example of the configuration of plant signal values ​​stored in a data storage unit. [Diagram 3] FIG. 4 is a diagram showing an example of the configuration of user input information stored in an input information storage unit; [Figure 4] FIG. 11 is a diagram showing an example of a process flow when control is performed. [Figure 5A] FIG. 4 is a diagram showing the configuration of an operation amount calculation means. [Figure 5B] FIG. 13 is a diagram showing an example of a processing flow of an operation amount calculation means. [Figure 6] FIG. 13 is a diagram showing an example of a processing flow of an IP type proportional correction unit of the manipulated variable calculation means. [Figure 7] FIG. 13 is a diagram showing an example of a processing flow of an IS type proportional correction unit of the manipulated variable calculation means. [Figure 8] FIG. 13 is a diagram showing an example of a process flow of an IP type integral correction unit of the manipulated variable calculation means. [Figure 9] FIG. 13 is a diagram showing an example of a process flow of an IS type integral correction unit of the manipulated variable calculation means. [Figure 10] FIG. 13 is a diagram showing an example of a process flow of a differential correction unit of the manipulated variable calculation means. [Figure 11] An example of a screen where the user inputs the necessary information via an I / O device. [Figure 12] An example of the control parameter input screen and the control result display screen. [Figure 13] FIG. 11 is a diagram showing an example of the configuration of a plant control system according to a second embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] The following describes examples of the present invention. EXAMPLES

[0013] (1) System Overview First Embodiment An overview of a plant control system according to a first embodiment of the present invention will be described with reference to FIG. The plant control system in FIG. 1 mainly comprises a model creation unit 100 and a control unit 110, which in turn comprise an input / output device 1, a data storage unit 2, an input information storage unit 3, an optimal transition model calculation means 4, an optimal transition model memory unit 5, an operation amount calculation means 6, and a plant 7.

[0014] In this system, a manipulated variable calculation means 6 calculates a manipulated variable based on an optimal transition value (the optimal transition value will be described later) output by an optimal transition model storage unit 5, the manipulated variable is received by the plant 7, and a signal value of the plant that has changed under the influence of the manipulated variable is input to the optimal transition model storage unit 5. By repeating this exchange, real-time control is performed to control the plant 7 to an optimal target. That is, an optimal transition model is created by the model creation unit 100 (described later), and the model is used in the control unit 110 to perform feedback control to always give an optimal control signal (manipulated variable) so that the transition value of the plant 7 (controlled object) becomes a target value. In addition, the creation process of the optimal transition model by the model creation unit 100 may be performed periodically or at a predetermined timing, such as when the plant as the controlled object starts operating or when the accuracy of the model is deteriorated.

[0015] With this configuration, the present plant control system is able to calculate appropriate manipulated variables even when the optimal transition values ​​become inappropriate due to the effects of disturbances or delays in the plant.

[0016] (2) Structure and function of each part Next, details of each part in this embodiment and their operations and functions will be described in detail below.

[0017] (2-1) Input / Output Device 1 The input / output device 1 is used by plant staff and the like to input various commands such as a processing start signal that starts processing in this plant control system, various setting values, and to check various data output from the plant controlled by this plant control system, and is composed of a display device (display), keyboard, mouse, etc. It may also be composed of a touch panel having an input function and an output function (display function).

[0018] (2-2) Data storage unit 2 A plurality of types of plant signal values ​​are stored in the data storage unit 2. Here, when the data storage unit 2 receives a processing start signal from the input / output device 1, it outputs the plant signal value to the optimal transition model calculation means 4. In addition, the plant signal values ​​observed by this plant control system are also accumulated in the data storage unit 2 in real time.

[0019] FIG. 2 is an example of past plant signal values ​​stored in the data storage unit 2. The first row of the table shown in FIG. 2 lists the name of the type of signal value, and the first column lists the acquisition time of the signal value. Then, each column after the first column stores time-series data of the signal value according to the name of the type of signal value. The signal values ​​stored here differ depending on the type of plant and the installation information of the sensor. Then, these signal values ​​are used when creating an optimal transition model in the optimal transition model calculation means 4 (the optimal transition model will be explained later).

[0020] (2-3) Input information storage unit 3 The input information storage unit 3 stores information input by the user via the input / output device 1. The information stored here is used by the optimal transition model calculation means 4 and the operation amount calculation means 6.

[0021] Fig. 3 is an example of the user input information stored in the input information storage unit 3. The first column of the table shown in Fig. 3 lists the item names, and the second column stores the information actually input by the user for each item. Among these, the signal to be used (E1) indicates various signal names in the plant used when the optimal transition destination calculation means 4 creates an optimal transition model. The operation signal (E2) indicates the name of the signal to be operated, and the controlled signal (E3) indicates the name of the signal to be controlled. The target value (E4) of the controlled signal indicates the target value of the controlled signal (E3). The control period (E5) indicates the control period in real-time control. In other words, it indicates the interval in seconds at which communication with the plant is performed.

[0022] These pieces of information will be specifically explained using a thermal power plant as an example. Thermal power plants use steam generated by combustion to turn turbines and generate electricity, but in order to keep the steam temperature constant at a target temperature, water is sprayed onto the generated steam. The amount of water sprayed is then adjusted to control the steam temperature.

[0023] In this case, the operation signal (E2) is the amount of injected water, the controlled amount signal (E3) is the steam temperature, and the target value (E4) is the target value of the controlled amount signal (E3) (target temperature of the steam temperature). The use signal (E1) indicates other signal values ​​including the steam temperature and the amount of injected water.

[0024] 3 are parameters used when calculating the manipulated variable by the manipulated variable calculation means 6, and are set (input) by a staff member or the like in the input / output device 1 based on past empirical rules, etc. A specific method of using them will be described later.

[0025] (2-4) Optimal transition model calculation means 4 The optimal transition model calculation means 4 receives past plant signal values ​​stored in the data storage unit 2 and user input information stored in the input information storage unit 3, and creates an optimal transition model. The optimal transition model is a model that predicts and outputs the state to which the controlled variable will transition in the future, using as input the signal value specified as the use signal in the user input information. How many hours into the future the predicted value is predicted depends on the control period. In other words, if the control period is one minute, the predicted value of the controlled variable one minute from now is output.

[0026] The optimal transition model may be created by an existing method such as the method disclosed in Patent Document 1 or a machine learning method. The optimal transition model may be stored in a table format or a specific formula with fixed parameters such as a neural network.

[0027] (2-5) Optimal transition model storage unit 5 The optimal transition model storage unit 5 stores the optimal transition model created by the optimal transition model calculation means 4. During control, the optimal transition model outputs an optimal transition value based on the real-time signal value output from the plant. Here, N hours of real-time signal values ​​are stored in the optimal transition model storage unit 5. N refers to the integral time included in the user input information. The optimal transition value refers to the value output by the optimal transition model. Specifically, it refers to a predicted value of the controlled signal of the plant, or some other indicated value.

[0028] (2-6) Manipulated amount calculation means 6 The operation amount calculation means 6 is composed of an IP type proportional compensation section 61, an IS type proportional compensation section 62, an IP type integral compensation section 63, an IS type integral compensation section 64, and a differential compensation section 65 shown in FIG. 1, and each compensation section calculates a compensation value. The IP type proportional compensation section 61 adds the compensation value to a reference operation amount, and the IS type proportional compensation section 62 to the differential compensation section 65 sequentially add the operation amount compensated by the compensation section arranged in the preceding stage to each calculated compensation value, thereby determining the final operation amount (optimum operation amount).

[0029] (2-7) Plant 7 The plant 7 has an operation terminal 71 where a user (such as an attendant) issues various commands and makes settings to operate and control the plant, and a measurement terminal 72 where various signal values ​​generated in association with the operation of the plant are measured.

[0030] (3) Overall flow of plant control The process flow when controlling a plant in real time is shown in Fig. 4. From now on, the flow of real time control will be explained using Fig. 1 and Fig. 4. In step S1, the optimal transition model storage unit 5 acquires a signal value measured from the plant 7. In step S2, the optimal transition model calculates an optimal transition value based on the signal value. The processes in steps S1 and S2 described above are performed in the optimal transition model storage unit 5.

[0031] In step S3, the manipulated variable calculation means 6 calculates an optimal manipulated variable for controlling the plant based on the optimal transition value and the reference manipulated variable, and outputs the calculated optimal manipulated variable to the plant 7. (The reference manipulated variable is a value that is defined in advance as a reference value of the manipulated variable corresponding to a combination of at least one or more signal values ​​based on the characteristics of the plant, and an appropriate value is adopted when the plant is designed. Alternatively, an arbitrary value can be set by a user (personnel) from the input / output device 1 after the plant is operational (in this case, the set value is stored in the input information storage unit 3 in the form of a table as shown in FIG. 3). A specific procedure for calculating the optimal manipulated variable in the manipulated variable calculation means 6 will be described later. The processing in step S3 described above is performed by the manipulated variable calculation means 6.

[0032] In step S4, the plant 7 acquires the optimal manipulated variable calculated by the manipulated variable calculation means 6, and the user operates the manipulated variable 71 based on the optimal manipulated variable. Then, the measurement terminal 72 measures a real-time signal value of the plant that has changed in response to the operation of the manipulated variable 71. In step S5, the real-time signal value is output to the input / output device 1 and the optimal transition model storage unit 5.

[0033] Then, sequential communication with the plant is performed through steps S1 to S5, and real-time control is implemented. This real-time control ends when the user sends an end instruction in the input / output device 1. A detailed example of the screen of the input / output device 1 will be described later.

[0034] (4) Calculation method for optimal operation amount Next, a detailed description will be given of the process of calculating the optimal manipulated variable in the manipulated variable calculation means 6. A detailed configuration of the manipulated variable calculation means 6 is shown in Fig. 5A, and a flow of the process of calculating the manipulated variable in the manipulated variable calculation means 6 is shown in Fig. 5B.

[0035] (4-1) Step SS1 In step SS1 shown in Fig. 5B, the reference manipulated variable 60, the optimal transition value (calculated in step S2 in Fig. 4) output from the optimal transition model storage unit 5, the real-time signal value, and user input information are acquired. The real-time signal value acquired here is a signal value set by the user as the use signal (E1) in the input / output device, and indicates an actual measurement value output from the plant 7. The user input information is a parameter used when calculating a correction value in each of the correction units 61 to 65, and specifically corresponds to the proportional gain Kp (E6) and proportional gain Ks (E7) to integral time N (E13) in Fig. 3.

[0036] (4-2) Step SS2 In step SS2, the proportional compensation value calculated in the IP type proportional compensation unit 61 is added to the reference manipulated variable to calculate the IP type proportional compensation reference manipulated variable. Fig. 6 shows a calculation process flow of the proportional compensation value in the IP type proportional compensation unit 61. In step SA1, the optimal transition value PV_Ideal of the controlled signal, the real-time controlled signal value PV_Real, and the proportional gain Kp are obtained. In step SA2, the proportional compensation value Cp is calculated using equation (1).

[0037]

number

[0038] Then, in step SA3, the proportional correction value Cp is added to the reference manipulated variable to calculate an IP type proportionally corrected reference manipulated variable.

[0039] (4-3) Step SS3 In SS3, the proportional compensation value calculated in the IS type proportional compensation unit 62 is added to the IP type proportional compensation base manipulated variable to calculate the IS type proportional compensation base manipulated variable.

[0040] 7 shows a calculation process flow of the proportional correction value in the IS type proportional correction unit 62. In step SB1, the optimal transition value PV_Ideal of the controlled signal, the target value SV of the controlled signal, and the proportional gain Ks are obtained. Next, in step SB2, the proportional correction value Cs is calculated using equation (2).

[0041]

number

[0042] Then, in step SB3, the proportional compensation value Cs is added to the IP type proportional compensation standard manipulated variable calculated in step SS2 to calculate an IS type proportional compensation standard manipulated variable.

[0043] (4-4) Step SS4 In step SS4, the integral correction value calculated by the IP type integral correction unit 63 is added to the IS type proportional correction reference manipulated variable to calculate the IP type integral correction reference manipulated variable.

[0044] 8 shows a process flow for calculating the integral correction value in the IP type integral correction unit 63. In step SC1, the optimum transition value PV_Ideal of the controlled signal for N seconds, the real-time controlled signal value PV_Real for N seconds, the integral saturation upper limit emax, and the integral saturation lower limit emin are obtained. The above N indicates the integral time N (E13) included in the user input information. In step SC2, the integral correction value Cip is calculated using equation (3). In step SC3, the value of Cip is determined using equation (4).

[0045] That is, the formula (4) serves to limit the value of Cip by conditional branching. Specifically, the value of Cip is limited so that it does not exceed the integral saturation upper limit emax and does not fall below the integral saturation lower limit emin.

[0046]

number

[0047]

number

[0048] In step SC4, the integral correction value Cip is added to the IS type proportional correction standard manipulated variable calculated in step SS3 to calculate an IP type integral correction standard manipulated variable.

[0049] (4-5) Step SS5 In step SS5, the integral correction value calculated by the IS type integral correction unit 64 is added to the IP type integral correction reference manipulated variable to calculate the IS type integral correction reference manipulated variable.

[0050] 9 shows a calculation process flow of the integral correction value in the IS type integral correction unit 64. In step SD1, the optimal transition value PV_Ideal of the controlled variable for N seconds, the target value SV of the controlled signal, the integral gain Kis, the integral saturation upper limit emax, and the integral saturation lower limit emin are obtained.

[0051] In step SD2, the integral correction value Cis is calculated using equation (5). In step SD3, the value of Cis is determined using equation (6). Equation (6) serves to limit the value of Cis by conditional branching, similar to the calculation of Cip.

[0052]

number

[0053]

number

[0054] In step SD4, the integral correction value Cis is added to the IP type integral correction reference operation amount calculated in step SS4 to calculate the IS type integral correction reference operation amount.

[0055] (4-6) Step SS6 In step SS6, the differential correction value calculated by the differential correction unit 65 is added to the IS type integral correction reference operation amount to calculate the final optimal operation amount.

[0056] Fig. 10 shows the calculation process flow of the differential correction value in the differential correction unit 65. In step SE1, the optimal transition value (dPV / dt)_Ideal of the change rate of the controlled signal, the real-time change rate (dPV / dt)_Real of the controlled signal, and the change rate gain Kr are obtained. Note that the optimal transition value of the change rate of the controlled signal and the real-time change rate of the controlled signal are calculated by the optimal transition model stored in the optimal transition model storage unit.

[0057] In step SE2, the integral correction value Cr is calculated using equation (7).

[0058]

Equation

[0059] In step SE3, the integral correction value Cr is added to the IS type integral correction reference operation amount calculated in step SS5 to calculate the final optimal operation amount.

[0060] (4-7) Effects of each correction unit Supplementary explanations of the effects of each correction unit (61 to 65) and their parameters are as follows. <IP type proportional correction unit 61> · When the response of the controlled variable to the change in the operation amount of the plant becomes smaller or larger than when learned for some reason (e.g., change in plant characteristics, etc.), there arises a problem that control cannot be performed well with the gain adjusted according to the conventional learning data. In such a case, when IP type proportional correction is used, it becomes possible to realize a control operation with an appropriate intensity corresponding to the change in the controlled variable by changing the control intensity so as to correct the deviation between the prediction and the actual measurement. As a result, it becomes possible to bring the controlled variable closer to the target value. · Depending on the magnitude of the gain, the user can appropriately adjust the responsiveness, making it possible to enhance the followability of the controlled variable to the target value.

[0061] <IS type proportional correction unit 62> · It has a function of bringing the controlled variable closer to the target value. · Depending on the magnitude of the gain, the user can appropriately adjust the responsiveness, making it possible to enhance the followability of the controlled variable to the target value. On the other hand, if the gain is too large, the force to approach becomes too strong, and the movement of the controlled variable may become oscillatory.

[0062] <IP type integral correction unit 63> · When the prediction of AI for the controlled variable and the actual behavior are consistently deviated due to long-period disturbances, etc., the gap can be filled and the controlled variable can be brought closer to the target value. · Depending on the magnitude of the gain, the user can appropriately adjust the responsiveness, making it possible to enhance the followability of the controlled variable to the target value.

[0063] <IS type integral correction unit 64> · When the AI prediction for the integrated variable has a deviation continuously for a certain period of time with respect to the target value of the controlled variable due to long-period disturbances, etc., the influence of the disturbance can be corrected and the target can be surely approached. · Depending on the magnitude of the gain, the force to resist disturbances is adjusted. Therefore, if the gain is too large, the controlled variable may exceed the target value, and it may take time to adjust the controlled variable to the target value.

[0064] <Differential correction unit 65> -Has the function of adjusting the speed at which the controlled quantity approaches the target value. The speed at which the controlled variable approaches the target value is adjusted depending on the gain. Therefore, if the gain is too large, it may take a long time for the controlled variable to approach the target value.

[0065] (4-8) Summary As described above, the manipulated variable calculation means 6 is composed of a plurality of correction units having user-adjustable parameters Kp, Ks, Kip, Kis, Kr, emax, emin, and N, which function together to enable appropriate control of the plant. Specifically, even if the plant state changes due to the effects of disturbances or delays, the user can set the parameters while observing the state and calculate appropriate manipulated variables.

[0066] That is, each of the correction units (61 to 65) can function effectively either alone or in any combination, and for example, depending on the plant state, it is possible to use only the IP type proportional correction unit 61 and the IS type integral correction unit 64, without using the other correction units. In this case, it is possible to set the proportional gain Kp of the IP type proportional correction unit 61 and the integral gain Kis of the IS type integral correction unit 64 to be larger than usual, and to set the gains of the other correction units to zero.

[0067] In addition, in the flow shown in FIG. 5, the IP type proportional compensation unit 61 adds the compensation value to the reference operation amount, and the IS type proportional compensation unit 62 to the differential compensation unit 65 sequentially add the operation amount compensated by the compensation unit arranged in the preceding stage to the compensation value calculated respectively. However, it is also possible to add up all the compensation values ​​calculated respectively by the compensation units 61 to 65 and then add the value to the reference operation amount.

[0068] (5) Display screen example Next, an example of the display screen of the input / output device 1 shown in Fig. 1 will be described. Fig. 11 and Fig. 12 show an example of a screen (Fig. 11) where the user inputs information required for executing the processes in the optimal transition model calculation means 4 and the manipulated variable calculation means 6, and a screen (Fig. 12) where the control results are output and gain adjustment is performed.

[0069] G1 in Fig. 11 displays a selection field for the user to select multiple types of signals output by the plant 7. The user can select one or multiple signals to be used for processing from the multiple signals in this G1. G2 displays an input field for the user to select the signal they want to operate. The user selects the signal they want to operate (operation signal) from among multiple signals. G3 displays an input field in which the user can select and input the signal value (controlled signal) that he / she wants to control. The user can select the controlled signal from multiple types of signals. G4 displays an input field for selecting and inputting a target value for the controlled signal. The user inputs an arbitrary value as the target value of the controlled signal. G5 displays an input field for the user to input the control period when performing control. The user inputs the control period. Note that this control period is input in seconds. G6 displays a control start button. After the user has input information into each of the above-mentioned input fields, the control system can start control by clicking the control start button. The G7 displays a stop control button. The user can stop real-time control by clicking the stop control button.

[0070] GG1 in Figure 12 displays a trend graph of a certain signal value. When you place the mouse cursor over this trend graph, the time and signal value at the point where the cursor is placed are displayed. The user selects the trend graph of the signal value to be displayed from a list of signal values ​​displayed in GG2. The vertical axis of the trend graph is displayed in percentages. GG3 displays an input field where the user can enter maximum and minimum values ​​to determine the display range of the vertical axis of the trend graph displayed in GG1. In other words, in terms of percentages on the vertical axis of GG1, the maximum value entered in GG3 corresponds to 100% and the minimum value corresponds to 0%.

[0071] Next, GG4 to GG11 display input fields for the user to input parameter values ​​used for adjusting the manipulated variable in the manipulated variable calculation means 6. In GG4, an input field for the user to input the proportional gain Kp used in the IP type proportional correction unit 61 is displayed. In GG5, an input field for the user to input the proportional gain Ks used in the IS type proportional correction unit 62 is displayed. In GG6, an input field is displayed for the user to input the integral gain Kip used in the IP type integral correction unit 63. In GG7, an input field is displayed for the user to input the integral gain Kis used in the IS type integral correction unit 64. GG8 displays an input field for the user to input the change rate gain Kr used in the differential correction unit 65. In GG9, an input field is displayed for the user to input the integral saturation upper limit value emax used in the IP type integral correction unit 63 and the IS type integral correction unit 64. The GG 10 displays an input field for the user to input the integral saturation lower limit emin used in the IP type integral corrector 63 and the IS type integral corrector 64 . In the GG 11, an input field is displayed for the user to input the integral time N used in the IP type integral correction unit 63 and the IS type integral correction unit 64. The user can set or adjust the parameters shown in GG4 to GG11 to optimal values ​​while checking the trend graph of the signal value shown in GG1. The optimal parameters mentioned here refer to a combination of parameters that allows the controlled signal to quickly settle to the target value.

[0072] As described above, according to this embodiment, the user can arbitrarily set a plurality of parameters while checking the plant state, thereby making it possible to calculate appropriate manipulated variables. This precise parameter adjustment makes it possible to appropriately control the plant even when the plant state changes due to disturbances or delays. EXAMPLES

[0073] In the first embodiment, the reference manipulated variable 60 is a preset value, but it may be difficult to set an optimal value. Therefore, in this embodiment, the optimal manipulated variable calculated one time ago (one control cycle ago) is used as the reference manipulated variable 60. Figure 13 shows the configuration of a plant control system in this embodiment. The difference from the first embodiment shown in Figure 1 is that an optimal manipulated variable storage means 8 for storing the optimal manipulated variable output from the manipulated variable calculation means 6 is newly provided.

[0074] That is, instead of inputting the reference manipulated variable 60 to the manipulated variable calculation means 6 in FIG. 1, the optimal manipulated variable output from the manipulated variable calculation means 6 is input to the plant 7 and also input to the optimal manipulated variable storage means 8, the value of which is stored, and used as the reference manipulated variable (reference manipulated variable 60A) when the manipulated variable calculation means 6 calculates the next optimal manipulated variable.

[0075] By employing this configuration, even when it is difficult to preset an optimal value for the reference manipulated variable, an appropriate value can be provided to the manipulated variable calculation means 6.

[0076] Furthermore, the optimum manipulated variable storage means 8 may store not only the optimum manipulated variable calculated one time ago (one control cycle ago), but also the optimum manipulated variable for a predetermined period of the past as time-series data, and input their average value to the manipulated variable calculation means 6. This makes it possible to reduce the influence of temporary large fluctuations in the optimum manipulated variable, and provide a more appropriate value to the manipulated variable calculation means 6.

[0077] Furthermore, the optimum manipulated variable storage means 8 may store past time-series data of the optimum manipulated variable together with the acquisition conditions (e.g., date, time period, weather, temperature, etc.), and when calculating a new optimum manipulated variable, calculate the reference manipulated variable using data acquired under similar conditions. This makes it possible to provide a more appropriate value to the manipulated variable calculation means 6. [Explanation of symbols]

[0078] 1: Input / output device 2: Data storage section 3: Input information storage section 4: Optimal transition model calculation method 5: Optimal transition model storage unit 5 6: Operation amount calculation means 7: Plant

Claims

1. A plant control system that calculates an optimal manipulated variable for making a value of a controlled variable of a controlled object plant follow its control target value, an optimal transition model storage unit in which an optimal transition model is stored, the optimal transition model being configured to calculate an optimal transition value, which is a value of a controlled quantity of a next state of the plant, based on a state transition probability model, based on a value of a controlled quantity obtained from a sensor installed in the plant; a manipulation amount calculation means for calculating the optimal manipulation amount by using a predetermined reference manipulation amount and an optimal transition value calculated by the optimal transition model; A plant control system comprising:

2. 2. The plant control system according to claim 1, the manipulated variable calculation means calculates a first proportional correction value based on a deviation between the optimal transition value and a measurement value of a controlled variable obtained by a sensor installed in the plant, and adds the first proportional correction value to the reference manipulated variable to calculate the optimal manipulated variable.

3. 2. The plant control system according to claim 1, the manipulated variable calculation means calculates a second proportional correction value based on a deviation between the optimal transition value and a predetermined target value of a controlled variable of the plant, and adds the second proportional correction value to the reference manipulated variable to calculate the optimal manipulated variable.

4. 2. The plant control system according to claim 1, the manipulated variable calculation means calculates an accumulated value within a predetermined time period for a deviation between the optimal transition value and a measurement value of a controlled variable obtained from a sensor installed in the plant, and calculates the optimal manipulated variable by adding a first integral correction value calculated based on the accumulated value to the reference manipulated variable.

5. 2. The plant control system according to claim 1, the manipulated variable calculation means calculates an accumulated value within a predetermined time period for a deviation between the optimal transition value and a predetermined target value of the controlled variable of the plant, and calculates the optimal manipulated variable by adding a second integral correction value calculated based on the accumulated value to the reference manipulated variable.

6. 2. The plant control system according to claim 1, the manipulated variable calculation means calculates a differential correction value based on a deviation between a time rate of change of the optimal transition value and a time rate of change of a measurement value of a controlled variable obtained from a sensor installed in the plant, and adds the differential correction value to the reference manipulated variable to calculate the optimal manipulated variable.

7. 2. The plant control system according to claim 1, The operation amount calculation means calculating a first proportional correction value based on a deviation between the optimal transition value and a measured value of a controlled quantity obtained from a sensor installed in the plant; calculating a second proportional correction value based on a deviation between the optimal transition value and a predetermined target value of a controlled variable of the plant; calculating an accumulated value within a predetermined time period for a deviation between the optimal transition value and a measured value of a controlled quantity obtained from a sensor installed in the plant, and calculating a first integral correction value based on the accumulated value; calculating an accumulated value within a predetermined time period for a deviation between the optimal transition value and a predetermined target value of a controlled variable of the plant, and calculating a second integral correction value based on the accumulated value; calculating a differential correction value based on a deviation between a time rate of change of the optimal transition value and a time rate of change of a measured value of a controlled quantity obtained from a sensor installed in the plant; a sum of the first proportional correction value, the second proportional correction value, the first integral correction value, the second integral correction value, and the differential correction value is added to the reference manipulated variable, thereby calculating the optimum manipulated variable.

8. A plant control system according to any one of claims 1 to 7, A storage means for storing the calculated optimum operation amount, A plant control system, comprising: a control unit for controlling a plant according to a first embodiment of the present invention;

9. 9. A plant control system according to claim 8, comprising: a control unit for controlling a plant according to a first embodiment of the present invention; a control unit for controlling a control operation based on the first embodiment of the present invention;

10. 8. A plant control system according to claim 7, comprising: an input / output device for inputting setting values ​​and commands for controlling the plant and outputting information regarding control results; a control unit for controlling a control amount of a plant based on the control amount calculated by the control amount calculation means, the control amount being calculated based on the control amount calculated by the control amount calculation means, and a control amount calculation unit for calculating a control amount of a plant based on the control amount calculated by the control amount calculation means.

11. A plant control method for calculating an optimal manipulated variable for making a value of a controlled variable of a controlled plant follow a control target value thereof, comprising the steps of: storing an optimal transition model for calculating an optimal transition value, which is a value of the controlled quantity of the next state of the plant, based on a state transition probability model, based on the value of the controlled quantity obtained from a sensor installed in the plant; Calculating the optimal manipulated variable using a predetermined reference manipulated variable and an optimal transition value calculated by the optimal transition model; A plant control method comprising: providing the calculated optimal operation amount to the plant.

Images