WO2020010629A1 - 过程控制预测模型调整方法、装置和过程控制器 - Google Patents

过程控制预测模型调整方法、装置和过程控制器 Download PDF

Info

Publication number
WO2020010629A1
WO2020010629A1 PCT/CN2018/095685 CN2018095685W WO2020010629A1 WO 2020010629 A1 WO2020010629 A1 WO 2020010629A1 CN 2018095685 W CN2018095685 W CN 2018095685W WO 2020010629 A1 WO2020010629 A1 WO 2020010629A1
Authority
WO
WIPO (PCT)
Prior art keywords
process control
prediction model
data
model
performance
Prior art date
Application number
PCT/CN2018/095685
Other languages
English (en)
French (fr)
Inventor
闻博
牛铸
范顺杰
Original Assignee
西门子股份公司
闻博
牛铸
范顺杰
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西门子股份公司, 闻博, 牛铸, 范顺杰 filed Critical 西门子股份公司
Priority to PCT/CN2018/095685 priority Critical patent/WO2020010629A1/zh
Priority to CN201880095609.8A priority patent/CN112424705A/zh
Priority to US17/259,672 priority patent/US11573542B2/en
Publication of WO2020010629A1 publication Critical patent/WO2020010629A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B17/00—Systems involving the use of models or simulators of said systems
    • G05B17/02—Systems involving the use of models or simulators of said systems electric
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/041—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a variable is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/047—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators the criterion being a time optimal performance criterion
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor

Definitions

  • the present application generally relates to the field of process control, and more particularly, to a method, an apparatus for adjusting a process control prediction model, and a process controller having the same.
  • Model Predictive Controller is the most popular Advanced Process Controller (APC), which has a dynamic model of the process.
  • APC Advanced Process Controller
  • MVs Manipulated Variables
  • the future change trend of the operating variables (Manipulated Variables, MVs) in the MPC controller is determined by performing optimizations aimed at maximizing the benefits of the plant under operating constraints. Because MPC controllers have the advantages of global optimization, MPC controllers have been widely used in industrial production.
  • the efficiency of the MPC controller depends heavily on the accuracy of the model. In recent years, many efforts have been made to make the system model identification process more efficient and accurate. However, no matter how accurate the model is, as conditions change, the model will always mismatch with the plant. Therefore, prediction errors will occur, and the efficiency of the MPC controller will be weakened, or even bring fluctuations to the system.
  • the present application provides a method, an apparatus for adjusting a process control prediction model, and a process controller having the same.
  • the performance of the MPC controller is continuously monitored and evaluated, and the process control prediction model adjustment is automatically triggered when the performance of the MPC controller is lower than the reference performance, thereby eliminating the need to perform retests to perform model Re-identification eliminates the above-mentioned mismatch, thereby eliminating the influence of fluctuations caused by increasing the excitation signal during re-testing.
  • a method for adjusting a process control prediction model includes: determining a process control prediction based on controlled variable (CV) data in process control data obtained through real-time monitoring. Whether the predicted performance of the model is lower than the reference performance; and when the predicted performance is lower than the reference performance, use the manipulated variable data (Manipulated Variables, MV) in the monitored process control data to predict the process control The model is adjusted.
  • CV controlled variable
  • determining whether the prediction performance of the process control prediction model is lower than the reference performance based on the controlled variable data in the process control data obtained through real-time monitoring may include: calculating a predetermined statistical period A standard deviation of a prediction error of the controlled variable data; and when the calculated standard deviation is greater than a reference threshold, determining that the prediction performance is lower than the reference performance.
  • adjusting the process control prediction model may include: obtaining the controlled variable after a specified period of time predicted by using the current process control prediction model under the given operation variable data Prediction data; obtaining actual data of the controlled variable after the specified period of time in the case of the given operating variable data; and based on the given operating variable data, the obtained predicted data of the controlled variable, and the obtained The actual data of the controlled variable adjusts the parameters of the current process control prediction model.
  • the process control data further includes Disturbance Variable (DV) data
  • the method may further include: when the predicted performance is lower than the reference performance, Use the manipulated variable data and disturbance variable data in the monitored process control data to determine a process control data statistical period suitable for the process control prediction model adjustment; and select the selected from the determined process control data statistical period Given manipulated variable data.
  • DV Disturbance Variable
  • the parameters of the process control prediction model include gain, steady state time, and dead time
  • adjusting parameters of the current process control prediction model may include: The gain of the current process control prediction model is adjusted.
  • the method may further include: using the adjusted process control prediction model to update a process control prediction model in a process controller.
  • the method may further include: adjusting the adjusted The process control prediction model performs prediction performance verification, and when the adjusted process control prediction model is verified that the degree of improvement in prediction performance does not exceed a predetermined threshold, the adjustment to the process control prediction model is re-executed, or the adjustment is performed during the adjustment.
  • the adjusted process control prediction model is used to update the process control prediction model in the process controller.
  • a device for adjusting a process control prediction model including: a model adjustment trigger judgment unit for determining based on controlled variable data in process control data obtained through real-time monitoring Whether the prediction performance of the process control prediction model is lower than the reference performance; and a model adjustment unit for using the operating variable data in the monitored process control data when the predicted performance is lower than the reference performance, The process control prediction model is adjusted.
  • the model adjustment trigger judging unit may include: a standard deviation calculation module for calculating a standard deviation of a prediction error of the controlled variable data in a predetermined statistical period; and model adjustment A triggering module is configured to determine that the predicted performance is lower than the reference performance when the calculated standard deviation is greater than a reference threshold.
  • the model adjustment unit may include: a controlled variable prediction data acquisition module for acquiring a designation predicted by using a current process control prediction model under given operating variable data Predicted data of the controlled variable after the period; actual data acquisition module of the controlled variable for acquiring the actual data of the controlled variable after the specified period under the condition of the given operational variable data; and a model parameter adjustment module, And used to adjust parameters of the current process control prediction model based on the given operating variable data, the obtained controlled variable prediction data, and the obtained controlled variable actual data.
  • the process control data further includes disturbance variable data
  • the apparatus may further include: an applicable data statistics period determining unit, configured to determine that the predicted performance is lower than the reference.
  • an applicable data statistics period determining unit configured to determine that the predicted performance is lower than the reference.
  • an operating variable data selection unit for selecting from The given operating variable data is selected in the determined process control data statistical period.
  • the parameters of the process control prediction model include gain, steady state time, and dead time
  • the model parameter adjustment module is configured to: Gain is adjusted.
  • the apparatus may further include: a model update unit, configured to use the adjusted process control prediction model to update a process control prediction model in a process controller.
  • a model update unit configured to use the adjusted process control prediction model to update a process control prediction model in a process controller.
  • the apparatus may further include a model verification unit configured to update the process control prediction model in the process controller before using the adjusted process control prediction model to update the process control prediction model in the process controller.
  • the adjusted process control prediction model performs prediction performance verification, and when the model verification unit verifies that the adjusted performance improvement degree of the adjusted process control prediction module does not exceed a predetermined threshold, the model adjustment unit re-executes For the adjustment of the process control prediction model, or when the model verification unit verifies that the adjusted performance improvement degree of the adjusted process control prediction model exceeds the predetermined threshold, the model update unit uses the adjusted The process control predictive model updates the process control predictive model in the process controller.
  • a process controller including: the device for adjusting a process control prediction model as described above; and a process control prediction model storage device for storing a process control prediction model.
  • a computing device including: one or more processors; and a memory coupled to the one or more processors, for storing instructions, when the instructions are described by the When executed by one or more processors, the processors are caused to execute the method for adjusting a process control prediction model as described above.
  • a non-transitory machine-readable storage medium which stores executable instructions that, when executed, cause the machine to execute the process control prediction model as described above.
  • the performance of the MPC controller is continuously monitored and evaluated, and the process control prediction model adjustment is automatically triggered when the performance of the MPC controller is lower than the reference performance, thereby eliminating the need Retesting is performed to re-identify the model to eliminate the above mismatches, thereby eliminating the effects of fluctuations introduced by increasing the stimulus signal during the retesting.
  • the model adjustment trigger judgment is performed by calculating the standard deviation of the prediction error of the controlled variable data within a predetermined statistical period, which can improve the accuracy of the model adjustment trigger judgment, thereby greatly reducing False triggering of model adjustments.
  • the process controller prediction model stored in the process controller can be updated by using the adjusted process control prediction model after performing the process control prediction model adjustment, so that the process controller can
  • the adjusted process control prediction model is used to execute the process prediction, thereby improving the prediction accuracy of the process controller.
  • the process of predicting the model of the process control is performed before the process control predicting model stored in the process controller is updated by using the adjusted process control predicting model, and the adjusted The above-mentioned update is performed only after the prediction performance improvement degree of the process control prediction model reaches a desired level, so that the process control prediction model in the process controller is optimized to meet user expectations.
  • FIG. 1 shows a schematic diagram of a model re-recognition process in the prior art
  • FIG. 2 is a schematic diagram showing an example of a binary random sequence signal
  • FIG. 3 is a block diagram showing a structure of a process controller according to an embodiment of the present application.
  • FIG. 4 shows a block diagram of an apparatus for adjusting a process control prediction model according to an embodiment of the present application
  • FIG. 5 is a structural block diagram showing an example of a model adjustment trigger judgment unit in FIG. 4;
  • FIG. 5 is a structural block diagram showing an example of a model adjustment trigger judgment unit in FIG. 4;
  • FIG. 6 is a block diagram showing an example of a model adjustment unit in FIG. 4;
  • FIG. 7 shows a flowchart of a method for adjusting a process control prediction model according to an embodiment of the present application
  • FIG. 8 illustrates a flowchart of an example of a process for adjusting a process control prediction model in FIG. 7;
  • FIG. 9 shows a flowchart of an example of the process for verifying the adjusted process control prediction model in FIG. 7;
  • FIG. 10 shows a performance comparison chart of a process controller adjusted by a process control prediction model according to the present application and a process controller in the prior art
  • FIG. 11 shows a block diagram of a computing device for adjusting a process control prediction model according to the present application.
  • processing unit DCS or PLC
  • S240 Adjust the process control prediction model based on the manipulated variable data, the obtained controlled variable prediction data, and the obtained controlled variable actual data.
  • the term “including” and variations thereof mean open terms, meaning “including but not limited to.”
  • the term “based on” means “based at least in part on.”
  • the terms “one embodiment” and “an embodiment” mean “at least one embodiment.”
  • the term “another embodiment” means “at least one other embodiment.”
  • the terms “first”, “second”, etc. may refer to different or the same objects. Other definitions can be included below, either explicitly or implicitly. Unless the context clearly indicates otherwise, the definition of a term is consistent throughout the specification.
  • FIG. 1 shows a schematic diagram of a model re-identification process in the prior art.
  • the identification device 10 sends an excitation signal 51 to the test device 20.
  • the test device 20 generates a test signal 52 based on the received excitation signal 51 and sends the test signal 52 to a process processing unit (DCS or PLC) 30.
  • the process processing unit 30 generates MV / DV / CV data 53 based on the received test signal 52 and sends the generated MV / DV / CV data 53 to the test device 20.
  • the test device 20 then sends this MV / DV / CV data 53 to the identification device 10.
  • the identification device 10 generates a new process control prediction model based on the MV / DV / CV data 53 and synchronizes the generated process control prediction model into the model file 40.
  • the excitation signal may be a step signal or a binary random sequence signal.
  • FIG. 2 is a schematic diagram showing an example of a binary random sequence signal.
  • the horizontal axis is a time axis, and its unit is minute
  • the vertical axis is a variation range of the manipulated variable, and its unit is consistent with the unit of the manipulated variable targeted.
  • FIG. 3 is a block diagram showing the structure of the process controller 1 according to the embodiment of the present application.
  • the process controller 1 includes a model adjustment device 100 and a process control prediction model storage device 200.
  • the model adjustment device 100 is used to adjust a process control prediction model.
  • the process control prediction model storage device 200 is configured to store a process control prediction model in the process controller for use by the process controller when performing process control prediction.
  • FIG. 4 shows a block diagram of a model adjustment apparatus 100 according to an embodiment of the present application.
  • the model adjustment device 100 includes a model adjustment trigger determination unit 110 and a model adjustment unit 130.
  • the model adjustment trigger judging unit 110 is configured to determine whether the prediction performance of the process control prediction model is lower than the reference performance based on the controlled variable data in the process control data obtained through real-time monitoring.
  • the process control data can be obtained by real-time monitoring of an automated control system or similar system located at a production site.
  • the model adjustment unit 130 is configured to adjust the process control prediction model using the operating variable data in the monitored process control data when the predicted performance is determined to be lower than the reference performance.
  • the process control data may include manipulated variable data, controlled variable data, and disturbance variable data.
  • Operating variables are variables that can be adjusted directly in process control.
  • the manipulated variable's adjusted value directly affects the controlled variable.
  • a controlled variable is a variable that needs to be adjusted to a specified value or region.
  • Disturbance variables are those variables that are not controllable and have an effect on the controlled variables.
  • the process controller monitors the values of the above three variables in real time, and uses the historical change trends of the above three variables to predict the future trend of the controlled variable, and gives the operating variables based on the future change trends. Adjust the value to keep the controlled variable stable. For example, suppose there is a heating tank, one feed and one discharge, and the temperature of the discharge is controlled by heating with steam. Then, the amount of steam is the operating variable, the discharge temperature is the controlled variable, and the feed temperature is the disturbance variable.
  • the prediction performance of the process control prediction model can be characterized by using the standard deviation of the prediction error of the controlled variable.
  • the characterization of the prediction performance of the process control prediction model is not limited to the standard deviation of the prediction error of the controlled variable.
  • the prediction performance of the process control prediction model may also use the controlled Variables are characterized by other statistical characteristics of the prediction error, such as maximum, minimum, average, median, etc.
  • the model adjustment trigger judgment unit 110 may include a standard deviation calculation module 111 and a model adjustment trigger judgment module 113, as shown in FIG. Show.
  • the standard variance calculation module 111 is configured to calculate a standard deviation of a prediction error of the controlled variable data in a predetermined statistical period.
  • n 1 to n.
  • n can be set based on specific conditions of the process processing unit.
  • the predicted value of the controlled variable and the actual value of the controlled variable of the MPC controller at each time point are distributed as Y m (t i ) and Y t (t i ).
  • the prediction error of the MPC controller is And the average prediction error over the entire statistical period is And the standard deviation of the prediction error is: Then, calculate the calculated standard deviation Compare with a set threshold (ie, reference performance). If the calculated standard deviation is greater than a reference threshold (ie, reference performance), the model adjustment trigger judgment module 113 determines that the prediction performance is lower than the reference performance. When the model adjustment trigger judgment module 113 determines that the prediction performance is lower than the reference performance, the operation control data in the process control data is used to adjust the process control prediction model.
  • the models in the MPC controller can usually be described using different types of models, such as a first-order model, a second-order model, or a slope model.
  • the model in the MPC controller is specifically described by which model described above, depending on the characteristics of the process to be controlled.
  • the key parameters (or transfer functions) of the above model include gain Gain, steady state time ⁇ , and dead time d.
  • gain Gain mismatch for example, the gain Gain mismatch due to changes in engineering load. Therefore, adjusting the process control prediction model in the MPC controller refers to adjusting the parameter gain Gain in the process control prediction model.
  • the manipulated variable input of the process controller can be expressed in the frequency domain as:
  • G t is the true gain value of the process control prediction model
  • E g is the gain error
  • the predicted output value of the process controller under the mismatch prediction model (ie, the current prediction model) is:
  • ⁇ is the steady-state time of the current process control prediction model
  • d is the dead time of the current process control prediction model
  • Mi is the adjusted manipulated variable input value
  • Y m (t) is the manipulated variable input
  • the gain error E g in the process control prediction model can be calculated by obtaining the values in formula (8), and then the gain parameters in the adjusted process control prediction model can be obtained, thereby achieving the process control prediction.
  • Model adjustments
  • FIG. 6 is a block diagram showing a configuration of an example of the model adjustment unit 140 in FIG. 4.
  • the model adjustment unit 140 may include a controlled variable prediction data acquisition module 141, a controlled variable actual data acquisition module 144, and a model parameter adjustment module 145.
  • the controlled variable prediction data acquisition module 141 is configured to obtain the controlled variable prediction data after a specified period of time predicted by using the current process control prediction model given the manipulated variable data.
  • the controlled variable actual data acquisition module 143 is configured to acquire the actual data of the controlled variable after the specified period of time has passed in the case of the given operation variable data.
  • the model parameter adjustment module 145 is configured to adjust parameters of the current process control prediction model based on the given operating variable data, the obtained controlled variable prediction data, and the obtained controlled variable actual data.
  • the model parameter adjustment module 145 calculates a gain error of a gain parameter of the process control prediction model based on the given operating variable data, the obtained controlled variable prediction data, and the obtained controlled variable actual data, and then, The gain parameter of the adjusted process control prediction model is determined using the gain error of the calculated gain parameter.
  • the process control data may further include disturbance variable data.
  • the model adjustment apparatus 100 may further include an applicable data statistical period determination unit 120 and an operation variable data selection unit 130.
  • the applicable data statistical period determination unit 120 uses the manipulated variable data and the disturbance variable data in the monitored process control data to determine process control data suitable for the process control prediction model adjustment. Statistical period.
  • the operation variable data selecting unit 130 selects the given operation variable data from the determined process control data statistical period.
  • the applicable data statistics period determination unit 120 evaluates the current state of the MPC controller based on the MV data and DV data in the monitored process control data. Specifically, calculate the standard deviations ⁇ MV and ⁇ DV of the MV data and DV data in the last n minutes, and then compare the calculated standard deviations ⁇ MV and ⁇ DV with a pre-created criterion with Compare.
  • pre-created guidelines with It is created by analyzing historical data, which can include CV data, MV data, DV data, and some other information, such as plant load, temperature, and so on.
  • the predetermined condition refers to a condition in which the current working condition belongs to a normal working condition.
  • the predetermined condition may be And Then, the operation variable data selection unit 130 selects the given operation variable data from the determined process control data statistical period, and then the model adjustment unit 140 performs a process control prediction based on the selected given operation variable data. Model adjustment. Otherwise, discard the current statistical period, that is, discard the data segment between the current time point and the starting point of the statistical period, and use the current point as the new starting point to continue statistics (for example, you can continue to count data for the predetermined period), The applicable data statistics period determination unit 120 determines whether the process control data statistics period is suitable for the process control prediction model adjustment based on the newly statistics data.
  • the model adjustment apparatus 100 may further include a model update unit 160.
  • the model updating unit 160 is configured to update the process control prediction model in the process control prediction model storage device after using the adjusted process control prediction model after obtaining the adjusted process control prediction model.
  • the model adjustment apparatus 100 may further include a model verification unit 150.
  • the model verification unit 150 is configured to perform prediction performance verification on the adjusted process control prediction model before using the adjusted process control prediction model to update the process control prediction model in the process controller.
  • the model verification unit 150 inputs the same operating variable values to the original process control prediction model and the adjusted process control prediction model, and calculates the respective process variables based on the original process control prediction model and the adjusted process control prediction module. The predicted value of the controlled variable, and then the prediction error of each controlled variable is obtained. Then, the model verification unit 150 calculates the standard deviation of the prediction error of the controlled variable under each model, and determines whether the prediction performance improvement of the adjusted process control prediction model exceeds a predetermined value based on the calculated standard deviations of the two prediction errors. Threshold? For example, by calculating Whether it is greater than a predetermined threshold to determine whether the predicted performance improvement exceeds a predetermined threshold. If it is determined that the prediction performance improvement of the adjusted process control prediction model exceeds a predetermined threshold, the model verification unit 150 determines that the prediction performance verification passes. Otherwise, the model verification unit 150 determines that the prediction performance verification fails.
  • the model adjustment unit 140 re-performs the adjustment for the process control prediction model.
  • the model update unit 160 updates the process control prediction model in the process controller by using the adjusted process control prediction model.
  • FIG. 7 shows a flowchart of a method for adjusting a process control prediction model according to an embodiment of the present application.
  • process control data is obtained through real-time monitoring. For example, real-time monitoring of an automated control system or similar system at a production site to obtain process control data. Then, in block S120, based on the controlled variable data in the monitored process control data, it is determined whether the prediction performance of the process control prediction model is lower than the reference performance. If the prediction performance is determined to be lower than the reference performance, the operation of block S130 is performed. If the predicted performance is determined to be not lower than the reference performance, it returns to block S110 and continues to monitor the automated control system. For operations of the blocks S110 and S120, refer to the operations of the model adjustment trigger determination unit 110 described above with reference to FIG. 4.
  • block S130 using the manipulated variable data and the disturbance variable data in the monitored process control data, a process control data statistical period suitable for the process control prediction model adjustment is determined.
  • a process control data statistical period suitable for the process control prediction model adjustment is determined. For the operation of block S130, see the operation of the applicable data statistics period determination unit 120 described above with reference to FIG. 4.
  • block S200 the process control prediction model is adjusted using the manipulated variable data in the monitored process control data.
  • the operation of the block S200 is referred to the operations of the manipulated variable data selection unit 130 and the model adjustment unit 140 described above with reference to FIG. 4.
  • FIG. 8 shows a flowchart of an example of a process for adjusting a process control prediction model in FIG. 7.
  • block S210 the operation variable data for the process control prediction model adjustment is selected from the determined process control data statistical period.
  • the operation of the block S210 is referred to the operation of the manipulated variable data selection unit 130 described above with reference to FIG. 4.
  • block S220 the controlled variable prediction data after the specified period predicted by the current process control prediction model in the case of the selected operating variable data is acquired.
  • the controlled variable prediction data acquisition module 141 described above with reference to FIG.
  • block S230 actual data of the controlled variable after the specified period of time is obtained in the case of the selected operating variable data.
  • the controlled variable actual data acquisition module 143 described above with reference to FIG.
  • block S240 parameters of the current process control prediction model are adjusted based on the manipulated variable data, the obtained controlled variable prediction data, and the obtained controlled variable actual data.
  • the model parameter adjustment module 145 described above with reference to FIG.
  • the adjusted process control prediction model is verified.
  • the operation of the block S300 is referred to the operation of the model verification unit 150 described above with reference to FIG. 4.
  • FIG. 9 shows a flowchart of an example of the process for verifying the adjusted process control prediction model in FIG. 7.
  • the model verification unit 140 inputs the same operating variable values to the original process control prediction model and the adjusted process control prediction model, and then in blocks S320-1 and S320-2, Based on the original process control prediction model and the adjusted process control prediction module, the predicted values of the controlled variables are calculated, and the prediction errors of the controlled variables are obtained. Then, in block S330, the standard deviation of the prediction error of the controlled variable under each model is calculated, and in block S340, the prediction of the adjusted process control prediction model is determined based on the calculated standard deviations of the two prediction errors Does the performance improvement exceed a predetermined threshold? For example, by calculating Whether it is greater than a predetermined threshold to determine whether the predicted performance improvement exceeds a predetermined threshold.
  • the adjusted process control prediction model is used to update the process control prediction model in the process controller, that is, the process control prediction model in the process control prediction model data storage device. For subsequent use by the process controller.
  • block S130 described above is optional. In some implementations of this application, the operation of block S130 may not be needed. Accordingly, the operation of block S210 is also not required. Similarly, the operations of blocks S300 and S400 described above are also optional operations.
  • FIG. 10 shows a performance comparison chart of a process controller adjusted by a process control prediction model according to the present application and a process controller in the prior art.
  • the scheme according to the present application is implemented using a 2-input (MV1 and MV2) 2-output (CV1 and CV2) system, where the gain value of the model between MV1 and CV1 has a high loss With (400%).
  • the gain value of the model between MV2 and CV1 is also set to have a small mismatch (20%).
  • the curve C2 has a very large overshoot after the two switches and has a relatively long steady-state time.
  • the above-mentioned overshoot and steady-state time are caused by model mismatch.
  • the mismatch model can be adjusted at the beginning of the first CV control point switch, and it has satisfactory performance at the second switch (ie, after adjustment of the overshoot control prediction model).
  • model adjustment device may be implemented by hardware, and may also be implemented by software or a combination of hardware and software.
  • FIG. 11 shows a block diagram of a computing device 1100 for adjusting a process control prediction model according to the present application.
  • the computing device 1100 may include a processor 1110 that executes one or more computer-readable instructions stored or encoded in a computer-readable storage medium (ie, the memory 1120) (ie, the above is in software form) Implementation element).
  • computer-executable instructions are stored in the memory 1120 which, when executed, cause one or more processors 1110 to determine a process control prediction based on controlled variable data in process control data obtained through real-time monitoring. Whether the predictive performance of the model is lower than the reference performance; and when the predictive performance is lower than the reference performance, use the manipulated variable data in the monitored process control data to adjust the process control prediction model.
  • a program product such as a non-transitory machine-readable medium.
  • the non-transitory machine-readable medium may have instructions (that is, the above-mentioned elements implemented in software form) that, when executed by a machine, cause the machine to execute various of the embodiments described above in connection with FIGS. Operation and function.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Feedback Control In General (AREA)

Abstract

一种用于调整过程控制预测模型的方法，包括：基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能；以及在预测性能被确定为低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据，对过程控制预测模型进行调整。利用该方法，可以无需执行重测试以对模型进行重识别来消除过程控制预测模型失配，从而消除由于在重测试期间增加激励信号而引入的波动影响。

Description

过程控制预测模型调整方法、装置和过程控制器 技术领域
本申请通常涉及过程控制领域，更具体地，涉及用于调整过程控制预测模型的方法、装置及具有该装置的过程控制器。
背景技术
在过程控制领域，模型预测控制器(Model Predictive Controller，MPC)是最流行的先进过程控制器(Advanced Process Controller，APC)，其具有过程的动态模型。MPC控制器中的操作变量(Manipulated Variable，MV)的未来变化趋势是通过进行目的为使得在操作约束下的工厂效益最大化的优化来确定的。由于MPC控制器具有全局化优化的优点，MPC控制器已经被广泛地应用于工业生产中。
然而，MPC控制器的效率严重地依赖于模型的准确性。近年来，已经进行了很多努力来使得系统模型识别过程更为高效和精确。然而，无论模型如何精确，随着条件变化，模型总会发生与工厂失配。因此，将会出现预测误差，并且MPC控制器的效率也将会被削弱，或者甚至会给系统带来波动。
在这些情况下，现有的MPC控制器会尝试通过执行重测试以对模型进行重识别来消除上述失配。然而，无论采用何种重识别方式(人工地或自动地)以及采用何种激励信号(阶跃测试序列或者二进制随机序列)，它们都将会影响正常生产过程。
发明内容
鉴于上述，本申请提供了一种用于调整过程控制预测模型的方法、装置及具有该装置的过程控制器。利用该方法及装置，通过持续监测和评估MPC控制器的性能，并且在MPC控制器的性能被评估为低于参考性能时自动触发过程控制预测模型调整，由此无需执行重测试以对模型进行重识 别来消除上述失配，从而消除由于在重测试期间增加激励信号而引入的波动影响。
根据本申请的一个方面，提供了一种用于调整过程控制预测模型的方法，包括：基于通过实时监测而获得的过程控制数据中的被控变量(Controlled Variable，CV)数据，确定过程控制预测模型的预测性能是否低于参考性能；以及在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据(Manipulated Variable，MV)，对所述过程控制预测模型进行调整。
可选地，在上述方面的一个示例中，基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能可以包括：计算预定统计时段内所述被控变量数据的预测误差的标准方差；以及在所计算出的标准方差大于参考阈值时，确定所述预测性能低于所述参考性能。
可选地，在上述方面的一个示例中，对所述过程控制预测模型进行调整可以包括：获取在给定操作变量数据的情况下使用当前过程控制预测模型所预测的指定时段后的被控变量预测数据；获取在所述给定操作变量数据的情况下经过所述指定时段后的被控变量实际数据；以及基于所述给定操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，对所述当前过程控制预测模型的参数进行调整。
可选地，在上述方面的一个示例中，所述过程控制数据还包括干扰变量(Disturbance Variable，DV)数据，以及所述方法还可以包括：在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于所述过程控制预测模型调整的过程控制数据统计时段；以及从所确定的过程控制数据统计时段中选择所述给定操作变量数据。
可选地，在上述方面的一个示例中，所述过程控制预测模型的参数包括增益、稳态时间和死区时间，以及对所述当前过程控制预测模型的参数进行调整可以包括：对所述当前过程控制预测模型的增益进行调整。
可选地，在上述方面的一个示例中，所述方法还可以包括：利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型。
可选地，在上述方面的一个示例中，在利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型之前，所述方法还可以包括：对所述调整后的过程控制预测模型进行预测性能验证，以及在所述调整后的过程控制预测模型被验证为预测性能改进度未超过预定阈值时，重新执行针对所述过程控制预测模型的调整，或者在所述调整后的过程控制预测模型被验证为预测性能改进度超过所述预定阈值时，利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型。
根据本申请的另一方面，提供了一种用于调整过程控制预测模型的装置，包括：模型调整触发判断单元，用于基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能；以及模型调整单元，用于在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据，对所述过程控制预测模型进行调整。
可选地，在上述方面的一个示例中，所述模型调整触发判断单元可以包括：标准方差计算模块，用于计算预定统计时段内所述被控变量数据的预测误差的标准方差；以及模型调整触发判断模块，用于在所计算出的标准方差大于参考阈值时，确定所述预测性能低于所述参考性能。
可选地，在上述方面的一个示例中，所述模型调整单元可以包括：被控变量预测数据获取模块，用于获取在给定操作变量数据的情况下使用当前过程控制预测模型所预测的指定时段后的被控变量预测数据；被控变量实际数据获取模块，用于获取在所述给定操作变量数据的情况下经过所述指定时段后的被控变量实际数据；以及模型参数调整模块，用于基于所述给定操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，对所述当前过程控制预测模型的参数进行调整。
可选地，在上述方面的一个示例中，所述过程控制数据还包括干扰变量数据，以及所述装置还可以包括：适用数据统计时段确定单元，用于在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于所述过程控制预测模型调整的过程控制数据统计时段；以及操作变量数据选择单元，用于从所确定的过程控制数据统计时段中选择所述给定操作变量数据。
可选地，在上述方面的一个示例中，所述过程控制预测模型的参数包括增益、稳态时间和死区时间，以及所述模型参数调整模块用于：对所述当前过程控制预测模型的增益进行调整。
可选地，在上述方面的一个示例中，所述装置还可以包括：模型更新单元，用于利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型。
可选地，在上述方面的一个示例中，所述装置还可以包括：模型验证单元，用于在利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型之前，对所述调整后的过程控制预测模型进行预测性能验证，以及在所述模型验证单元验证为所述调整后的过程控制预测模块的预测性能改进度未超过预定阈值时，所述模型调整单元重新执行针对所述过程控制预测模型的调整，或者在所述模型验证单元验证为所述调整后的过程控制预测模型的预测性能改进度超过所述预定阈值时，所述模型更新单元利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型。
根据本申请的另一方面，提供了一种过程控制器，包括：如上所述的用于调整过程控制预测模型的装置；以及过程控制预测模型存储装置，用于存储过程控制预测模型。
根据本申请的另一方面，提供了一种计算设备，包括：一个或多个处理器；以及与所述一个或多个处理器耦合的存储器，用于存储指令，当所述指令被所述一个或多个处理器执行时，使得所述处理器执行如上所述的用于调整过程控制预测模型的方法。
根据本申请的另一方面，提供一种非暂时性机器可读存储介质，其存储有可执行指令，所述指令当被执行时使得所述机器执行如上所述的用于调整过程控制预测模型的方法。
利用根据本申请的方法、装置及过程控制器，通过持续监测和评估MPC控制器的性能，并且在MPC控制器的性能被评估为低于参考性能时自动触发过程控制预测模型调整，由此无需执行重测试以对模型进行重识别来消除上述失配，从而消除由于在重测试期间增加激励信号而引入的波动影响。
利用根据本申请的方法、装置及过程控制器，通过计算预定统计时段 内被控变量数据的预测误差的标准方差来进行模型调整触发判断，可以提高模型调整触发判断的准确性，由此大大降低模型调整的错误触发情形。
利用根据本申请的方法、装置及过程控制器，通过在预测性能低于参考性能时，使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于过程控制预测模型调整的过程控制数据统计时段，并且从所确定的过程控制数据统计时段中选择给定操作变量数据，从而可以使得过程控制预测模型调整中所使用的操作变量数据更加准确，由此提高过程控制预测模型调整的准确率。
利用根据本申请的方法、装置及过程控制器，通过在执行过程控制预测模型调整后，利用调整后的过程控制预测模型来更新过程控制器中存储的过程控制预测模型，可以使得过程控制器能够利用调整后的过程控制预测模型来执行过程预测，从而提高过程控制器的预测准确率。
利用根据本申请的方法、装置及过程控制器，通过在利用调整后的过程控制预测模型来更新过程控制器中存储的过程控制预测模型之前执行过程控制预测模型验证过程，可以保证在调整后的过程控制预测模型的预测性能改进度达到期望水平后才进行上述更新，从而使得过程控制器中的过程控制预测模型优化满足用户的期望要求。
附图说明
通过参照下面的附图，可以实现对于本公开内容的本质和优点的进一步理解。在附图中，类似组件或特征可以具有相同的附图标记。
图1示出了现有技术中的模型再识别过程的示意图；
图2示出了二进制随机序列信号的一个示例的示意图；
图3示出了根据本申请的实施例的过程控制器的结构的方框图；
图4示出了根据本申请的实施例的用于调整过程控制预测模型的装置的方框图；
图5示出了图4中的模型调整触发判断单元的示例的结构方框图；
图6示出了图4中的模型调整单元的示例的结构方框图；
图7示出了根据本申请的实施例的用于调整过程控制预测模型的方法的流程图；
图8示出了图7中的用于对过程控制预测模型进行调整的过程的一个示例的流程图；
图9示出了图7中的用于对调整后的过程控制预测模型进行验证的过程的一个示例的流程图；
图10示出了根据本申请的经过过程控制预测模型调整后的过程控制器与现有技术中的过程控制器的性能比较图；和
图11示出了根据本申请的用于调整过程控制预测模型的计算设备的方框图。
附图标记
10  识别设备
20  测试设备
30  处理单元(DCS或PLC)
40  模型文件
51  激励信号
52  测试信号
53  MV/DV/CV数据
1   过程控制器
100 模型调整装置
200  过程控制预测模型存储装置
110  模型调整触发判断单元
120  适用数据统计时段确定单元
130  操作变量数据选择单元
140  模型调整单元
150  模型验证单元
160  模型更新单元
111  标准方差计算模块
113  模型调整触发判断模块
141  被控变量预测数据获取模块
143  被控变量实际数据获取模块
145  模型参数调整模块
S110  实时监测过程控制数据
S120  判断过程控制预测模型的预测性能是否低于参考性能
S130  使用过程控制数据中的操作变量数据和干扰变量数据，确定适合用于所述过程控制预测模型调整的过程控制数据统计时段
S200  过程控制预测模型调整过程
S300  过程控制预测模型验证过程
S400  过程控制预测模型更新过程
S210  选择操作变量数据
S220  获取被控变量预测数据
S230  获取被控变量实际数据
S240  基于操作变量数据、所获取的被控变量预测数据、所获取的被控变量实际数据，对过程控制预测模型进行调整
S310  选择操作变量输入
S320-1  基于原始过程控制预测模型生成被控变量预测值
S320-2  基于调整后的过程控制预测模型生成被控变量预测值
S330  计算各个模型下的被控变量预测误差的标准方差
S340  确定调整后的过程控制预测模型的性能改进是否超过预定阈值？
1100    计算设备
1110    一个或多个处理器
1120    存储器
具体实施方式
现在将参考示例实施方式讨论本文描述的主题。应该理解，讨论这些实施方式只是为了使得本领域技术人员能够更好地理解从而实现本文描述的主题，并非是对权利要求书中所阐述的保护范围、适用性或者示例的限制。可以在不脱离本公开内容的保护范围的情况下，对所讨论的元素的功能和排列进行改变。各个示例可以根据需要，省略、替代或者添加各种过程或组件。例如，所描述的方法可以按照与所描述的顺序不同的顺序来执 行，以及各个步骤可以被添加、省略或者组合。另外，相对一些示例所描述的特征在其它例子中也可以进行组合。
如本文中使用的，术语“包括”及其变型表示开放的术语，含义是“包括但不限于”。术语“基于”表示“至少部分地基于”。术语“一个实施例”和“一实施例”表示“至少一个实施例”。术语“另一个实施例”表示“至少一个其他实施例”。术语“第一”、“第二”等可以指代不同的或相同的对象。下面可以包括其他的定义，无论是明确的还是隐含的。除非上下文中明确地指明，否则一个术语的定义在整个说明书中是一致的。
图1示出了现有技术中的模型再识别过程的示意图。如图1所示，识别设备10将激励信号51发送给测试设备20。测试设备20基于所接收的激励信号51来生成测试信号52，并将测试信号52发送给过程处理单元(DCS或PLC)30。过程处理单元30基于所接收的测试信号52生成MV/DV/CV数据53，并将所生成的MV/DV/CV数据53发送给测试设备20。测试设备20随后将该MV/DV/CV数据53发送给识别设备10。识别设备10基于该MV/DV/CV数据53生成新的过程控制预测模型，并将所生成的过程控制预测模型同步到模型文件40中。
在本申请中，激励信号可以是阶跃信号或者二进制随机序列信号。图2示出了二进制随机序列信号的一个示例的示意图。在图2中示出的示例中，横轴是时间轴，其单位为分钟，以及纵轴是操作变量的变化幅度，其单位与所针对的操作变量的单位一致。
在图1中示出的模型再识别过程中，需要执行重测试以对模型进行重识别来完成过程控制预测模型调整，并且在重测试期间要向过程处理单元添加二进制随机序列信号，从而会给正常生产过程造成波动影响。此外，为了减少识别设备10输出的激励信号所引入的波动，需要减少再识别机制的激活次数，从而不能实现对过程控制预测模型的实时更新。
图3示出了根据本申请的实施例的过程控制器1的结构的方框图。如图3所示，过程控制器1包括模型调整装置100和过程控制预测模型存储装置200。模型调整装置100用于对过程控制预测模型进行调整。过程控制预测模型存储装置200用于存储过程控制器中的过程控制预测模型，以供过程控制器在进行过程控制预测时使用。
图4示出了根据本申请的实施例的模型调整装置100的方框图。如图4所示，模型调整装置100包括模型调整触发判断单元110和模型调整单元130。
模型调整触发判断单元110用于基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能。例如，所述过程控制数据可以通过对位于生产站点处的自动化控制系统或类似系统进行实时监测而获得。模型调整单元130用于在所述预测性能被确定为低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据，对所述过程控制预测模型进行调整。
在本申请中，过程控制数据可以包括操作变量数据、被控变量数据以及干扰变量数据。操作变量指的是在过程控制中可以直接对其进行调节的变量。操作变量的调节值会直接影响被控变量。被控变量指的是需要被调节至指定数值或区域的变量。干扰变量指的是那些不可控并且会对被控变量产生影响的变量。在本申请中，过程控制器实时监测上述三个变量的取值，并且使用上述三个变量的历史变化趋势来预测出被控变量的未来趋势，并且根据该未来变化趋势来给出操作变量的调节值，从而保持被控变量的稳定。例如，假设存在一个加热罐，一路进料以及一路出料，并且通过蒸汽来加热控制出料温度。那么，蒸汽量就是操作变量，出料温度是被控变量，以及进料温度是干扰变量。
在本申请的一个示例中，过程控制预测模型的预测性能可以利用被控变量的预测误差的标准方差来表征。本领域技术人员可以理解，对过程控制预测模型的预测性能的表征不限于被控变量的预测误差的标准方差，在本申请的另一示例中，过程控制预测模型的预测性能也可以利用被控变量的预测误差的其它统计特征来表征，比如，最大值、最小值、平均值、中位数等。
在过程控制预测模型的预测性能可以利用被控变量的预测误差的标准方差来表征的情况下，模型调整触发判断单元110可以包括标准方差计算模块111和模型调整触发判断模块113，如图5所示。
具体地，标准方差计算模块111用于计算预定统计时段内所述被控变量数据的预测误差的标准方差。
例如，假设t i是时间点，以及统计时段为[t 1，t 2，...,t (n-(n-i)),...,t n]，i＝1到n。这里，n可以基于过程处理单元的具体条件来设定。每个时间点的MPC控制器的被控变量预测值和被控变量实际值分布为Y m(t i)和Y t(t i)。
由此，MPC控制器的预测误差为
Figure PCTCN2018095685-appb-000001
并且在整个统计时段内的平均预测误差为
Figure PCTCN2018095685-appb-000002
以及预测误差的标准方差为：
Figure PCTCN2018095685-appb-000003
然后，将所计算出的标准方差
Figure PCTCN2018095685-appb-000004
与所设定的阈值(即，参考性能)进行比较。如果所计算出的计算出的标准方差大于参考阈值(即，参考性能)，则模型调整触发判断模块113确定预测性能低于参考性能。在模型调整触发判断模块113确定预测性能低于参考性能时，使用过程控制数据中的操作变量数据，对过程控制预测模型进行调整。
在过程控制中，MPC控制器中的模型通常可以利用1阶模型、2阶模型或斜坡模型等不同种类的模型来描述。MPC控制器中的模型具体采用上述哪种模型描述，取决于待控制的过程的特性。上述模型的关键参数(或者，传递函数)包括增益Gain、稳态时间τ和死区时间d。在上述三个关键参数中，最普遍的模型失配情形是增益Gain失配，例如，由于工程负载变化而导致的增益Gain失配。因此，针对MPC控制器中的过程控制预测模型进行的调整是指对过程控制预测模型中的参数增益Gain进行调整。
下面针对过程控制预测模型中的参数增益Gain进行调整的原理进行说明。
针对MPC控制器，假设T是控制时段，在每个控制时段中，操作变量MV的值变化一次，例如，将操作变量MV变化一个阶跃步长(即，阶跃幅度)后得到新的操作变量MV的输入值Mi，其中i是控制时段的序号。由此，过程控制器的操作变量输入在频域中可以被表示为：
Figure PCTCN2018095685-appb-000005
假设过程控制器中的过程控制预测模型是具有时延的1阶模型，则传递函数为：
Figure PCTCN2018095685-appb-000006
如果过程控制预测模型存在增益误差E g，则
Gain＝G t+E g     公式(3)
其中，G t是过程控制预测模型的真实增益值，E g是增益误差。
过程控制器在失配预测模型(即，当前预测模型)下的预测输出值为：
Figure PCTCN2018095685-appb-000007
并且，实际输出值为：
Figure PCTCN2018095685-appb-000008
将上述公式(4)和(5)从频域转换到时域，则可以得到：
Figure PCTCN2018095685-appb-000009
从上面的公式(6)和(7)可以得出，
Figure PCTCN2018095685-appb-000010
从上面可以看出，τ是当前过程控制预测模型的稳态时间，d是当前过程控制预测模型的死区时间，Mi是调整后的操作变量输入值，Y m(t)是在操作变量输入Mi的情况下基于当前过程控制预测模型预测出的预测输出值，以及Y t(t)是在操作变量输入Mi的情况下经过时间t后的实际输出值。由此，可以通过得到公式(8)中的各个值来计算出过程控制预测模型中的增益误差E g，进而得出调整后的过程控制预测模型中的增益参数，由此实现针对过程控制预测模型的调整。
图6示出了图4中的模型调整单元140的示例的结构方框图。如图6所示，模型调整单元140可以包括被控变量预测数据获取模块141、被控变量实际数据获取模块144和模型参数调整模块145。
被控变量预测数据获取模块141用于获取在给定操作变量数据的情况 下使用当前过程控制预测模型所预测的指定时段后的被控变量预测数据。被控变量实际数据获取模块143用于获取在所述给定操作变量数据的情况下经过所述指定时段后的被控变量实际数据。模型参数调整模块145用于基于所述给定操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，对所述当前过程控制预测模型的参数进行调整。具体地，模型参数调整模块145基于所述给定操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，计算出过程控制预测模型的增益参数的增益误差，然后，利用所计算出的增益参数的增益误差，确定出调整后的过程控制预测模型的增益参数。
此外，在本申请的另一示例中，过程控制数据还可以包括干扰变量数据。模型调整装置100还可以包括适用数据统计时段确定单元120和操作变量数据选择单元130。在预测性能被确定为低于参考性能时，适用数据统计时段确定单元120使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于过程控制预测模型调整的过程控制数据统计时段。然后，操作变量数据选择单元130从所确定出的过程控制数据统计时段中选择出所述给定操作变量数据。
具体地，适用数据统计时段确定单元120基于所监测到的过程控制数据中的MV数据和DV数据，对MPC控制器的当前状态进行评估。具体地，计算最近n分钟内的MV数据和DV数据的标准方差σ MV和σ DV，然后将所计算出的标准方差σ MV和σ DV与预先创建的准则
Figure PCTCN2018095685-appb-000011
和
Figure PCTCN2018095685-appb-000012
进行比较。这里，预先创建的准则
Figure PCTCN2018095685-appb-000013
和
Figure PCTCN2018095685-appb-000014
是通过对历史数据进行分析后创建的，所述历史数据可以包含CV数据、MV数据、DV数据以及一些其他信息，比如工厂负载、温度等。
如果上述比较满足预定条件，所述预定条件是指当前工况属于正常工况的条件，例如，所述预定条件可以是
Figure PCTCN2018095685-appb-000015
且
Figure PCTCN2018095685-appb-000016
然后，操作变量数据选择单元130从所确定出的过程控制数据统计时段中选择出所述给定操作变量数据，并且随后模型调整单元140基于所选择出的给定操作变量数据来执行过程控制预测模型调整。否则，丢弃当前统计时段，即，丢弃当前时点与统计时段起始点之间的数据段，并且把当前点作为新的起始点，继续进行统计(例如，可以继续统计预定时段的数据)，并 且适用数据统计时段确定单元120基于新统计的数据来来确定该过程控制数据统计时段是否适合用于过程控制预测模型调整。
模型调整装置100还可以包括模型更新单元160。模型更新单元160用于在如上获得调整后的过程控制预测模型后，利用所述调整后的过程控制预测模型来更新过程控制预测模型存储设备中的过程控制预测模型。
此外，可选地，模型调整装置100还可以包括模型验证单元150。模型验证单元150用于在利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型之前，对所述调整后的过程控制预测模型进行预测性能验证。
具体地，模型验证单元150向原始的过程控制预测模型和调整后的过程控制预测模型输入相同的操作变量值，并且基于原始的过程控制预测模型和调整后的过程控制预测模块计算出各自的被控变量预测值，进而得出各自的被控变量预测误差。然后，模型验证单元150计算出各个模型下的被控变量预测误差的标准方差，并且基于所计算出的两个预测误差的标准方差来判断调整后的过程控制预测模型的预测性能改进是否超过预定阈值？例如，通过计算
Figure PCTCN2018095685-appb-000017
是否大于预定阈值来确定预测性能改进是否超过预定阈值。如果判断出调整后的过程控制预测模型的预测性能改进超过预定阈值，则模型验证单元150确定预测性能验证通过。否则，模型验证单元150确定预测性能验证不通过。
在模型验证单元150确定预测性能验证不通过时，模型调整单元140重新执行针对过程控制预测模型的调整。在模型验证单元150确定预测性能验证通过时，模型更新单元160利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型。
上面参照图3到图6对根据本申请的用于调整过程控制预测模型的装置和具有该装置的过程控制器进行描述，下面结合图7到图9描述根据本申请的用于调整过程控制预测模型的方法。
图7示出了根据本申请的实施例的用于调整过程控制预测模型的方法的流程图。
如图7所示，在块S110，通过实时监测以获得过程控制数据。例如，对位于生产站点处的自动化控制系统或类似系统进行实时监测以获得过程 控制数据。然后，在块S120，基于所监测到的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能。如果预测性能被确定为低于参考性能，则执行块S130的操作。如果预测性能被确定为不低于参考性能，则返回到块S110，继续监测自动化控制系统。块S110和块S120的操作参见上面参照图4描述的模型调整触发判断单元110的操作。
接着，在块S130中，使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于过程控制预测模型调整的过程控制数据统计时段。块S130的操作参见上面参照图4描述的适用数据统计时段确定单元120的操作。
然后，在块S200，使用所监测到的过程控制数据中的操作变量数据，对所述过程控制预测模型进行调整。块S200的操作参见上面参照图4描述的操作变量数据选择单元130和模型调整单元140的操作。
图8示出了图7中的用于对过程控制预测模型进行调整的过程的一个示例的流程图。
如图8中所示，在块S210，从所确定的过程控制数据统计时段中选择用于所述过程控制预测模型调整的操作变量数据。块S210的操作参见上面参照图4描述的操作变量数据选择单元130的操作。
在块S220，获取在所选择的操作变量数据的情况下使用当前过程控制预测模型所预测的指定时段后的被控变量预测数据。块S220的操作参见上面参照图5描述的被控变量预测数据获取模块141的操作。
接着，在块S230，获取在所选择的操作变量数据的情况下经过所述指定时段后的被控变量实际数据。块S230的操作参见上面参照图5描述的被控变量实际数据获取模块143的操作。
然后，在块S240，基于所述操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，对所述当前过程控制预测模型的参数进行调整。块S240的操作参见上面参照图5描述的模型参数调整模块145的操作。
在如上对过程控制预测模型进行调整后，在块S300，对调整后的过程控制预测模型进行验证。块S300的操作参见上面参照图4描述的模型验证 单元150的操作。
图9示出了图7中的用于对调整后的过程控制预测模型进行验证的过程的一个示例的流程图。
如图9中所示，在块S310，模型验证单元140向原始的过程控制预测模型和调整后的过程控制预测模型输入相同的操作变量值，随后在块S320-1和块S320-2中，分别基于原始的过程控制预测模型和调整后的过程控制预测模块计算出各自的被控变量预测值，进而得出各自的被控变量预测误差。然后，在块S330，计算出各个模型下的被控变量预测误差的标准方差，并且在块S340中，基于所计算出的两个预测误差的标准方差来判断调整后的过程控制预测模型的预测性能改进是否超过预定阈值？例如，通过计算
Figure PCTCN2018095685-appb-000018
是否大于预定阈值来确定预测性能改进是否超过预定阈值。如果判断出调整后的过程控制预测模型的预测性能改进超过预定阈值，则确定预测性能验证通过，流程进行到块S400的操作。否则，确定预测性能验证不通过，流程返回到块S130的操作，继续执行过程控制预测模型调整。
在如上确定预测性能验证通过后，在块S400，利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型，即过程控制预测模型数据存储设备中的过程控制预测模型，以供过程控制器后续使用。
这里要说明的是，上面描述的块S130的操作是可选操作。在本申请的一些实现方式中，可以不需要块S130的操作。相应地，块S210的操作也不需要。同样，上面描述的块S300和S400的操作也是可选操作。
图10示出了根据本申请的经过过程控制预测模型调整后的过程控制器与现有技术中的过程控制器的性能比较图。
在图10中示出的性能模拟中，根据本申请的方案利用2输入(MV1和MV2)2输出(CV1和CV2)系统实现，其中MV1和CV1之间的模型的增益值存在较高的失配(400％)。此外，为了模拟实际环境同时证明根据本申请的方案的鲁棒性，MV2和CV1之间的模型的增益值也被设置为存在较小的失配(20％)。
上述模拟是在Simulink/Matlab中实现的。CV1的控制点在时间点0处被从0切换到100，以及在时间点75处被从100切换到200。图10中示出 了控制器性能比较，其中，曲线C1表示经过过程控制预测模型调整后的过程控制器的性能曲线，以及曲线C2表示现有技术中的过程控制器的性能曲线。
从图10中可以看出，曲线C2在两个切换之后具有非常大的超调(overshoot)并且具有相当长的稳态时间，上述超调和稳态时间是由于模型失配造成的。然而，在曲线C1中，可以在第一次CV控制点切换的开始处调整该失配模型，并且在第二次切换处(即，经过超调控制预测模型调整后)具有满意的性能。
如上参照图3到图10，对根据本申请的用于调整过程控制预测模型的方法、模型调整装置以及具有该装置的过程控制器的实施例进行了描述。上面的模型调整装置可以采用硬件实现，也可以采用软件或者硬件和软件的组合来实现。
在本申请中，模型调整装置100可以利用计算设备实现。图11示出了根据本申请的用于调整过程控制预测模型的计算设备1100的方框图。根据一个实施例，计算设备1100可以包括处理器1110，处理器1110执行在计算机可读存储介质(即，存储器1120)中存储或编码的一个或多个计算机可读指令(即，上述以软件形式实现的元素)。
在一个实施例中，在存储器1120中存储计算机可执行指令，其当执行时使得一个或多个处理器1110：基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能；以及在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据，对所述过程控制预测模型进行调整。
应该理解，在存储器1120中存储的计算机可执行指令当执行时使得一个或多个处理器1110进行本申请的各个实施例中以上结合图3-10描述的各种操作和功能。
根据一个实施例，提供了一种比如非暂时性机器可读介质的程序产品。非暂时性机器可读介质可以具有指令(即，上述以软件形式实现的元素)，该指令当被机器执行时，使得机器执行本申请的各个实施例中以上结合图3-10描述的各种操作和功能。
上面结合附图阐述的具体实施方式描述了示例性实施例，但并不表示 可以实现的或者落入权利要求书的保护范围的所有实施例。在整个本说明书中使用的术语“示例性”意味着“用作示例、实例或例示”，并不意味着比其它实施例“优选”或“具有优势”。出于提供对所描述技术的理解的目的，具体实施方式包括具体细节。然而，可以在没有这些具体细节的情况下实施这些技术。在一些实例中，为了避免对所描述的实施例的概念造成难以理解，公知的结构和装置以框图形式示出。
本公开内容的上述描述被提供来使得本领域任何普通技术人员能够实现或者使用本公开内容。对于本领域普通技术人员来说，对本公开内容进行的各种修改是显而易见的，并且，也可以在不脱离本公开内容的保护范围的情况下，将本文所定义的一般性原理应用于其它变型。因此，本公开内容并不限于本文所描述的示例和设计，而是与符合本文公开的原理和新颖性特征的最广范围相一致。

Claims (17)

  1. 一种用于调整过程控制预测模型的方法，包括：
    基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能；以及
    在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据，对所述过程控制预测模型进行调整。
  2. 如权利要求1所述的方法，其中，基于通过实时监测而获得的实时过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能包括：
    计算预定统计时段内所述被控变量数据的预测误差的标准方差；以及
    在所计算出的标准方差大于参考阈值时，确定所述预测性能低于所述参考性能。
  3. 如权利要求1或2所述的方法，其中，对所述过程控制预测模型进行调整包括：
    获取在给定操作变量数据的情况下使用当前过程控制预测模型所预测的指定时段后的被控变量预测数据；
    获取在所述给定操作变量数据的情况下经过所述指定时段后的被控变量实际数据；以及
    基于所述给定操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，对所述当前过程控制预测模型的参数进行调整。
  4. 如权利要求3所述的方法，其中，所述过程控制数据还包括干扰变量数据，以及所述方法还包括：
    在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于所述过程控制预测模型调整的过程控制数据统计时段；以及
    从所确定的过程控制数据统计时段中选择所述给定操作变量数据。
  5. 如权利要求4所述的方法，其中，所述过程控制预测模型的参数包括增益、稳态时间和死区时间，以及
    对所述当前过程控制预测模型的参数进行调整包括：对所述当前过程控制预测模型的增益进行调整。
  6. 如权利要求1所述的方法，还包括：
    利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型。
  7. 如权利要求6所述的方法，在利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型之前，所述方法还包括：
    对所述调整后的过程控制预测模型进行预测性能验证，以及
    在所述调整后的过程控制预测模型被验证为预测性能改进度未超过预定阈值时，重新执行针对所述过程控制预测模型的调整，或者
    在所述调整后的过程控制预测模型被验证为预测性能改进度超过所述预定阈值时，利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型。
  8. 一种用于调整过程控制预测模型的装置(100)，包括：
    模型调整触发判断单元(110)，用于基于通过实时监测而获得的过程控制数据中的被控变量数据，确定过程控制预测模型的预测性能是否低于参考性能；以及
    模型调整单元(140)，用于在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据，对所述过程控制预测模型进行调整。
  9. 如权利要求8所述的装置(100)，其中，所述模型调整触发判断单元(110)包括：
    标准方差计算模块(111)，用于计算预定统计时段内所述被控变量数 据的预测误差的标准方差；以及
    模型调整触发判断模块(113)，用于在所计算出的标准方差大于参考阈值时，确定所述预测性能低于所述参考性能。
  10. 如权利要求8或9所述的装置(100)，其中，所述模型调整单元(140)包括：
    被控变量预测数据获取模块(141)，用于获取在给定操作变量数据的情况下使用当前过程控制预测模型所预测的指定时段后的被控变量预测数据；
    被控变量实际数据获取模块(143)，用于获取在所述给定操作变量数据的情况下经过所述指定时段后的被控变量实际数据；以及
    模型参数调整模块(145)，用于基于所述给定操作变量数据、所获取的被控变量预测数据和所获取的被控变量实际数据，对所述当前过程控制预测模型的参数进行调整。
  11. 如权利要求10所述的装置(100)，其中，所述过程控制数据还包括干扰变量数据，以及所述装置(100)还包括：
    适用数据统计时段确定单元(120)，用于在所述预测性能低于所述参考性能时，使用所监测到的过程控制数据中的操作变量数据和干扰变量数据，确定适合用于所述过程控制预测模型调整的过程控制数据统计时段；以及
    操作变量数据选择单元(130)，用于从所确定的过程控制数据统计时段中选择所述给定操作变量数据。
  12. 如权利要求11所述的装置(100)，其中，所述过程控制预测模型的参数包括增益、稳态时间和死区时间，以及
    所述模型参数调整模块(145)用于：对所述当前过程控制预测模型的增益进行调整。
  13. 如权利要求8所述的装置(100)，还包括：
    模型更新单元(160)，用于利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型。
  14. 如权利要求13所述的装置(100)，还包括：
    模型验证单元(150)，用于在利用所述调整后的过程控制预测模型来更新过程控制器中的过程控制预测模型之前，对所述调整后的过程控制预测模型进行预测性能验证，以及
    在所述模型验证单元(150)验证为所述调整后的过程控制预测模块的预测性能改进度未超过预定阈值时，所述模型调整单元(140)重新执行针对所述过程控制预测模型的调整，或者
    在所述模型验证单元(150)验证为所述调整后的过程控制预测模型的预测性能改进度超过所述预定阈值时，所述模型更新单元(160)利用所述调整后的过程控制预测模型更新所述过程控制器中的过程控制预测模型。
  15. 一种过程控制器(10)，包括：
    如权利要求8到14中任一所述的装置(100)；以及
    过程控制预测模型存储装置(200)，用于存储过程控制预测模型。
  16. 一种计算设备(1100)，包括：
    一个或多个处理器(1110)；以及
    与所述一个或多个处理器(1110)耦合的存储器(1120)，用于存储指令，当所述指令被所述一个或多个处理器执行时，使得所述处理器执行如权利要求1到7中任一所述的方法。
  17. 一种非暂时性机器可读存储介质，其存储有可执行指令，所述指令当被执行时使得所述机器执行如权利要求1到7中任一所述的方法。
PCT/CN2018/095685 2018-07-13 2018-07-13 过程控制预测模型调整方法、装置和过程控制器 WO2020010629A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
PCT/CN2018/095685 WO2020010629A1 (zh) 2018-07-13 2018-07-13 过程控制预测模型调整方法、装置和过程控制器
CN201880095609.8A CN112424705A (zh) 2018-07-13 2018-07-13 过程控制预测模型调整方法、装置和过程控制器
US17/259,672 US11573542B2 (en) 2018-07-13 2018-07-13 Method and apparatus for adjusting process control prediction model and process controller

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/095685 WO2020010629A1 (zh) 2018-07-13 2018-07-13 过程控制预测模型调整方法、装置和过程控制器

Publications (1)

Publication Number Publication Date
WO2020010629A1 true WO2020010629A1 (zh) 2020-01-16

Family

ID=69143186

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/095685 WO2020010629A1 (zh) 2018-07-13 2018-07-13 过程控制预测模型调整方法、装置和过程控制器

Country Status (3)

Country Link
US (1) US11573542B2 (zh)
CN (1) CN112424705A (zh)
WO (1) WO2020010629A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113765957A (zh) * 2020-06-04 2021-12-07 华为技术有限公司 一种模型更新方法及装置
WO2023240546A1 (zh) * 2022-06-16 2023-12-21 北京小米移动软件有限公司 模型监测方法、装置、设备及介质
CN118444984A (zh) * 2024-04-18 2024-08-06 北京微核芯科技有限公司 处理器前端指令读取队列的实现方法及装置
WO2024169634A1 (zh) * 2023-02-16 2024-08-22 大唐移动通信设备有限公司 信息传输方法、装置及设备

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3521949B1 (de) * 2018-02-01 2021-01-13 Siemens Aktiengesellschaft Vorrichtung zum simulieren einer gesteuerten maschine oder anlage sowie verfahren
CN111539550B (zh) * 2020-03-13 2023-08-01 远景智能国际私人投资有限公司 光伏阵列工作状态的确定方法、装置、设备及存储介质
CN118249683B (zh) * 2024-05-21 2024-10-15 国家电投集团江西电力有限公司新昌发电分公司 基于自主可控算法的火电机组智能励磁控制方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101349893A (zh) * 2007-07-18 2009-01-21 太极光控制软件（北京）有限公司 自适应模型预测控制装置
CN101587328A (zh) * 2008-05-23 2009-11-25 朱豫才 非线性过程动态模型辨识装置
CN104793489A (zh) * 2015-02-12 2015-07-22 浙江大学 一种基于收敛深度控制的嵌入式mpc控制方法
CN105700358A (zh) * 2016-03-14 2016-06-22 华中科技大学 一种带漂移干扰的模型预测控制器的建模质量监控方法

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7024253B2 (en) * 2000-08-21 2006-04-04 Honeywell International Inc. Auto-tuning controller using loop-shaping
US7050863B2 (en) * 2002-09-11 2006-05-23 Fisher-Rosemount Systems, Inc. Integrated model predictive control and optimization within a process control system
US7451004B2 (en) 2005-09-30 2008-11-11 Fisher-Rosemount Systems, Inc. On-line adaptive model predictive control in a process control system
US8606544B2 (en) * 2006-07-25 2013-12-10 Fisher-Rosemount Systems, Inc. Methods and systems for detecting deviation of a process variable from expected values
US8185217B2 (en) * 2008-01-31 2012-05-22 Fisher-Rosemount Systems, Inc. Robust adaptive model predictive controller with tuning to compensate for model mismatch
CN102985884B (zh) * 2010-04-19 2015-11-25 Abb研究有限公司 用于更新模型预测控制器中模型的方法和系统
US10222769B2 (en) * 2012-10-12 2019-03-05 Emerson Process Management Power & Water Solutions, Inc. Method for determining and tuning process characteristic parameters using a simulation system
CN103092074B (zh) 2012-12-30 2015-09-09 重庆邮电大学 半导体先进过程控制的参数优化控制方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101349893A (zh) * 2007-07-18 2009-01-21 太极光控制软件（北京）有限公司 自适应模型预测控制装置
CN101587328A (zh) * 2008-05-23 2009-11-25 朱豫才 非线性过程动态模型辨识装置
CN104793489A (zh) * 2015-02-12 2015-07-22 浙江大学 一种基于收敛深度控制的嵌入式mpc控制方法
CN105700358A (zh) * 2016-03-14 2016-06-22 华中科技大学 一种带漂移干扰的模型预测控制器的建模质量监控方法

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113765957A (zh) * 2020-06-04 2021-12-07 华为技术有限公司 一种模型更新方法及装置
CN113765957B (zh) * 2020-06-04 2022-09-16 华为技术有限公司 一种模型更新方法及装置
WO2023240546A1 (zh) * 2022-06-16 2023-12-21 北京小米移动软件有限公司 模型监测方法、装置、设备及介质
WO2024169634A1 (zh) * 2023-02-16 2024-08-22 大唐移动通信设备有限公司 信息传输方法、装置及设备
CN118444984A (zh) * 2024-04-18 2024-08-06 北京微核芯科技有限公司 处理器前端指令读取队列的实现方法及装置

Also Published As

Publication number Publication date
US11573542B2 (en) 2023-02-07
US20210232106A1 (en) 2021-07-29
CN112424705A (zh) 2021-02-26

Similar Documents

Publication Publication Date Title
WO2020010629A1 (zh) 过程控制预测模型调整方法、装置和过程控制器
JP6513015B2 (ja) 機械の動作を制御する方法、および機械の動作を反復的に制御する制御システム
JP6193891B2 (ja) 調整可能なトレードオフ係数を用いて非干渉性の閉ループステップ試験を行う装置および方法
CN108151253B (zh) 一种变风量空调送风温度自动补偿方法
JP2016100009A5 (zh)
JP2015507300A5 (zh)
US20090132064A1 (en) Apparatuses, Systems, and Methods Utilizing Adaptive Control
NZ587705A (en) Method for the computer-assisted learning of control and/or regulation of a technical system
JP2011003186A5 (zh)
WO2019051963A1 (zh) 基于全回路重构仿真的工业控制回路性能评价方法及装置
JP6927446B1 (ja) 制御装置、制御方法及びプログラム
KR20200063338A (ko) 학습을 통한 파라미터 개선 기반의 예측 시스템 및 방법
WO2016092872A1 (ja) 制御装置、そのプログラム、プラント制御方法
CN105989374A (zh) 一种在线训练模型的方法和设备
JP6147687B2 (ja) 制御パラメータ調整装置、および制御パラメータ調整プログラム
JP7014330B1 (ja) 制御装置、制御方法、及びプログラム
JP2006302078A (ja) 制御対象モデル生成装置および生成方法
Jiang et al. Robust approximate dynamic programming and global stabilization with nonlinear dynamic uncertainties
CN108628288B (zh) 一种用于一阶时滞组合积分控制系统的性能评价方法
JP2006127079A (ja) 制御対象モデル生成装置および生成方法
JP2018528511A (ja) 生産システムにおける出力効率の最適化
JP7045857B2 (ja) モデル予測制御アプリケーションにおける最高性能値に関する優れた性能のためのシステム及び方法
Romero et al. Event-Based PI controller with adaptive thresholds
CN112488319B (zh) 一种具有自适应配置生成器的调参方法和系统
JP7043261B2 (ja) モデル予測コントローラ及び推定器の機器点検のためのシステム及び方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18925886

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18925886

Country of ref document: EP

Kind code of ref document: A1