JP6464899B2 - Multivariable predictive control system - Google Patents

Description

  The present invention relates to a multivariable predictive control system using a process model, and more particularly to a multivariable predictive control system capable of comparing a predicted value and an actual value retroactively to a past time point.

  A multivariable predictive control system for controlling a real process by predicting a change in process value using a multivariable process model corresponding to the dynamic characteristics of the real process has been put into practical use. FIG. 9 is a block diagram illustrating a configuration example of a conventional multivariable predictive control system 40.

  As shown in this figure, the multivariable predictive control system 40 is a system that predicts changes in process values and controls the actual process 500, and includes a multivariable predictive control unit 400 and an operation / monitoring unit 460. .

  The multivariable prediction control unit 400 is a block that performs process value prediction processing and control of the actual process 500 by setting various operation amounts and the like in the actual process 500 and acquiring various measurement values from the actual process 500. It is. The operation / monitoring unit 460 is a block for a user to perform an operation on the multivariable prediction control unit 400 or monitor the actual process 500, and includes an input device, a display device, and the like.

  The multivariable prediction control unit 400 includes a controller unit 410, a process actual measurement value DB 420, and a trend display control unit 430. The controller unit 410 includes a process model 411 designed for the actual process 500. The process model 411 is, for example, a multivariable model in which the relationship between the operation amount, disturbance, control amount, and the like of the actual process 500 is represented by a transfer function such as a primary delay system, a secondary delay system, and a dead time system.

  The controller unit 410 controls the actual process 500 using the process model 411 and records various process measured values acquired from the actual process 500 in the process measured value DB 420.

  The prediction unit 412 uses the process model 411 and parameters corresponding to the operation amount, measurement value, and the like of the actual process 500 to predict the process value of how the process value changes from the present time. The process value to be predicted can be specified by the user via the operation / monitoring unit 460.

  The trend display control unit 430 displays the trend of the process actual value recorded up to the current time recorded in the process actual measurement value DB 420 and the trend of the process value after the current time predicted by the prediction unit 412 via the operation / monitoring unit 460. To display.

  FIG. 10 shows an example of the current prediction trend screen on which the actual measurement value trend and the prediction value trend are displayed. On the current predicted trend screen, the latest parameter, that is, a predicted value based on the state of the actual process 500 at the current time is displayed. Below, the time of the state used for prediction shall be called prediction reference time. In this example, the prediction reference time is the current time.

Japanese Patent Laying-Open No. 2005-092584

  If the multivariable predictive control system 400 is controlling a certain real process 500, if it is possible to evaluate its controllability, that is, determine whether the control state is good or bad, the process model 411 It is useful as an opportunity to correct or adjust the actual process 500. However, evaluation of controllability is generally not easy and no method has been established.

  Accordingly, an object of the present invention is to make it easy to evaluate controllability in a multivariable predictive control system using a process model.

In order to solve the above problems, a multivariable predictive control system of the present invention predicts a change in a process value of the real process using a process model corresponding to the real process, and controls the real process. Then, when an instruction at a past time point is received, a trend of a process predicted value predicted at the past time point and a trend of a process actual measurement value measured after the past time point from the past time point as a starting point. A trend display control unit for superimposing display is provided.
Here, the trend display control unit displays the trend of the measured process value and a cursor indicating the current time in an initial state, and accepts the movement of the cursor in the past direction, thereby indicating the past time point. Can be accepted.
The trend display control unit can obtain the process predicted value at the past time point with reference to a database in which the process predicted value predicted at a predetermined prediction timing is recorded in association with the time.

  According to the present invention, controllability can be easily evaluated in a multivariable predictive control system using a process model.

It is a block diagram which shows the structural example of the multivariable prediction control system of this embodiment. It is a flowchart explaining operation | movement of a estimation part. It is a flowchart explaining operation | movement of a trend display control part. It is a figure explaining the example of the trend screen of this embodiment. It is a figure explaining the example of the trend screen of this embodiment. It is a figure explaining the example of the trend screen of this embodiment. It is a figure explaining the example of the trend screen of this embodiment. It is a figure explaining the example of the trend screen of this embodiment. It is a block diagram which shows the structural example of the conventional multivariable prediction control system. It is a figure which shows the example of the present prediction trend screen on which the actual value trend and the prediction value trend were displayed.

  Embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a multivariable predictive control system 10 of the present embodiment. As shown in this figure, the multivariable predictive control system 10 is a system that predicts a change in process value using a process model and controls the actual process 200. The multivariable predictive control unit 100 and the operation / monitoring unit 160 And.

  The multivariable prediction control unit 100 is a block that performs process value prediction processing and control of the actual process 200 by setting various manipulated variables in the actual process 200 and acquiring various measurement values from the actual process 200. . The operation / monitoring unit 160 is a block for the user to perform operations on the multivariable prediction control unit 100 or monitor the actual process 200, and includes an input device, a display device, and the like.

  The multivariable prediction control unit 100 includes a controller unit 110, a process actual measurement value DB 120, a process prediction value DB 130, and a trend display control unit 140. The controller unit 110 includes a process model 111 designed to correspond to the actual process 200. The process model 111 is a multivariable model in which the relationship between the operation amount, disturbance, control amount, and the like of the actual process 200 is represented by a transfer function such as a first-order lag system, a second-order lag system, and a dead time system.

  The controller unit 110 controls the actual process 200 using the process model 111, and records various process measured values acquired from the actual process 200 in the process measured value DB 120.

  The prediction unit 112 uses the process model 111 and parameters corresponding to the operation amount, measurement value, and the like of the actual process 200 to predict the process value of how the process value changes from the present. When the prediction unit 112 predicts the process value, the prediction unit 112 adds information on a time that is a reference for the prediction and records it in the process prediction value DB 130.

  The trend display control unit 140 sends the trend of the process actual value recorded up to the current time recorded in the process actual measurement value DB 120 and the trend of the process value after the current time predicted by the prediction unit 112 via the operation / monitoring unit 160. To display.

  Here, evaluation of controllability in the multivariable predictive control system 10 of the present embodiment will be described. Since the multivariable predictive control system 10 controls the actual process 200 using the process model 111, the controllability is better when the difference (model error) between the process model 111 and the actual process 200 is smaller. Yes, it can be said that the larger the model error, the worse the condition.

  Since the multivariable predictive control system 10 predicts a process value using the process model 111, it grasps a model error by comparing the predicted value of the process value with the actual process value, and evaluates the controllability. It is thought that you can. That is, it can be said that the controllability is good if the predicted value trend at a certain point coincides with the trend of the actually measured value obtained after a certain period of time has elapsed since that point. On the other hand, if they do not match, it can be seen that sufficient controllability is not obtained.

  For this reason, in addition to the conventional current prediction trend screen that displays the actual value trend up to the current time and the prediction value trend after the current time, the trend display control unit 140 includes the prediction value trend at the past time point and the past time point. The trend screen on which the actual measured value trend is superimposed can be displayed via the operation / monitoring unit 160.

  Thereby, the user can compare the predicted value trend at the past time point with the actually measured value trend from the past time point, and can easily evaluate the controllability of the multivariable predictive control system 10. The past time point can be set by the user, and the predicted value trend at the past time point can be acquired by referring to the process predicted value DB 130.

  Next, a characteristic operation of the multivariable predictive control system 10 of the present embodiment will be described. First, the process value prediction performed by the prediction unit 112 and the recording operation of the process predicted value in the process predicted value DB 130 will be described with reference to the flowchart of FIG. This process is performed for each process value to be predicted.

  When the predetermined prediction timing comes (S101: Yes), the prediction unit 112 predicts a change in process value using various parameters corresponding to the actual process 200 (S102). The predicted value is expressed in a time series from the time when the prediction is performed to the future in a predetermined period.

  Note that the prediction timing can be set for each process value. For example, if the process value changes frequently, the prediction is performed at a short time interval, and if the process value does not change much, the prediction is performed at a long time interval.

  Then, the predicted values obtained in time series are recorded in the process predicted value DB 130 (S103). At this time, the time when the prediction is performed, that is, the prediction reference time is recorded in association with each other.

  The prediction unit 112 repeats the above processing. Thereby, the prediction value trend at that time is recorded in the process prediction value DB 130 for each prediction timing.

  Next, the trend screen display operation performed by the trend display control unit 140 will be described with reference to the flowchart of FIG. This process starts when the trend display control unit 140 receives a trend screen display instruction from the user via the operation / monitoring unit 160 (S201).

  When the trend display control unit 140 receives an instruction to display a trend screen (S201: Yes), the trend display control unit 140 displays a trend screen as shown in FIG.

  As shown in FIG. 4, in the initial state, the trend screen displays a trend of measured values for a period before the current time, and a predicted value using the current time as a prediction reference time for a period before the current time. Displays the trend of. A cursor C is displayed at the current time, indicating the boundary between the measured value trend and the predicted value trend.

  On the trend screen, as shown in FIG. 5, the user can move the cursor C displayed at the current time in the past direction. Furthermore, the user can switch on the past predicted value trend display button S with the cursor C moved in the past direction.

  When the trend display control unit 140 receives the movement of the cursor C from the user (S203: Yes) and receives the ON of the past predicted value trend display button S (S204: Yes), the trend display control unit 140 determines the position of the moved cursor C. It acquires (S205), and acquires the predicted value trend at the past time indicated by the position from the process predicted value DB 130 (S206). At this time, the movement unit of the cursor C may be adjusted to the interval of the prediction timing, or a predicted value trend at the prediction timing closest to the time indicated by the position of the cursor C may be acquired.

  Then, the trend screen is redrawn using the acquired predicted value trend at the past time (S207). FIG. 6 shows an example of a redrawn trend screen. As shown in the figure, the cursor C has been moved to a past time by the user, and on the trend screen, a predicted value trend with the past time as a prediction reference time and an actual value trend up to the current time are superimposed. Has been displayed.

  Thereby, the predicted value trend at a certain point in the past can be compared with the actual trend, and the model error can be visually grasped. Therefore, the controllability of the multivariable predictive control system 10 can be easily evaluated. It becomes like this.

  At this time, if the past predicted value exists after the current time, the past predicted value trend may be displayed over the past period and the future period as shown in FIG.

  On the trend screen, the past predicted value trend display button S can be turned off to hide the past predicted value trend. When the user moves the cursor C again without ending the trend screen display (S208: No) (S203: Yes), the processing after the processing (S204) is repeated, and the time indicated by the cursor C after the movement is displayed. What is necessary is just to display the predicted value trend and actual value trend which were made into prediction reference time. When the user ends the trend screen display (S208: Yes), this process ends.

  In the display example described above, one process value is displayed. However, as shown in FIG. 8, a plurality of process values can be displayed on the trend screen.

  As described above, according to the multivariable predictive control system 10 of the present embodiment, it becomes possible to compare a predicted value trend and a measured value trend at a past time point. It is possible to easily evaluate whether or not good controllability is obtained.

DESCRIPTION OF SYMBOLS 10 ... Multivariable prediction control system, 40 ... Multivariable prediction control system, 100 ... Multivariable prediction control part, 110 ... Controller part, 111 ... Process model, 112 ... Prediction part, 120 ... Process measurement value DB, 130 ... Process prediction Value DB, 140 ... Trend display control unit, 160 ... Operation / monitoring unit, 200 ... Real process

Claims (3)

A multi-variable predictive control system that predicts a change in the process value of the actual process using a process model corresponding to the actual process and controls the actual process,
When receiving an instruction of a past time, the starting point of the past time point, said the trend of the process predicted value predicted by past time, to superimpose and trend of process actual value measured after the time of the last A multivariable predictive control system comprising a trend display control unit. The trend display control unit displays a trend of the process actual measurement value and a cursor indicating a current time in an initial state, and accepts an instruction at the past time point by accepting movement of the cursor in a past direction. The multivariable predictive control system according to claim 1 .   The trend display control unit acquires the process predicted value at the past time point with reference to a database in which the process predicted value predicted at a predetermined prediction timing is recorded in association with the time. The multivariable predictive control system according to 1 or 2.