IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0698991
(2010-02-02)
|
등록번호 |
US-8200346
(2012-06-12)
|
발명자
/ 주소 |
|
출원인 / 주소 |
- Fisher-Rosemount Systems, Inc.
|
대리인 / 주소 |
Marshall, Gerstein & Borun LLP
|
인용정보 |
피인용 횟수 :
32 인용 특허 :
7 |
초록
▼
An MPC controller technique integrates feedback control performance better than methods commonly used today in MPC type controllers, resulting in an MPC controller that performs better than traditional MPC techniques in the presence of process model mismatch. In particular, MPC controller performanc
An MPC controller technique integrates feedback control performance better than methods commonly used today in MPC type controllers, resulting in an MPC controller that performs better than traditional MPC techniques in the presence of process model mismatch. In particular, MPC controller performance is enhanced by adding a tunable integration block to the MPC controller that develops an integral component indicative of the prediction or other control error, and adds this component to the output of an MPC controller algorithm to provide for faster or better control in the presence of model mismatch, which is the ultimate reason for the prediction error in the first place. This technique enables the MPC controller to react more quickly and to provide better set point change and load disturbance performance in the presence of model mismatch, without decreasing the robustness of the MPC controller.
대표청구항
▼
1. A model based process controller for use in controlling a process, comprising: a controller input to receive one or more process measurements indicative of a controlled variable within the process;a controller output to provide one or more control signals for controlling a manipulated process var
1. A model based process controller for use in controlling a process, comprising: a controller input to receive one or more process measurements indicative of a controlled variable within the process;a controller output to provide one or more control signals for controlling a manipulated process variable of the process;a process model unit including a process model that models the operation of a process, wherein the process model unit produces a predicted process variable value based on the process model;a control unit that uses the predicted process variable value and a process variable set point to generate a preliminary control signal;an error unit that uses the predicted process variable value to generate an error signal;an integrator that integrates the error signal to produce an integrated error signal; anda combiner unit coupled to the integrator and to the control unit to combine the integrated error signal with the preliminary control signal to produce a final control signal for use in controlling the manipulated process variable of the process. 2. The model based process controller of claim 1, wherein the error unit determines an error between the predicted process variable value and a set point for the controlled variable and wherein the controller unit uses the error to produce the preliminary control signal. 3. The model based process controller of claim 1, wherein the error unit determines an error between the predicted process variable value and a measured value of the controlled variable to determine the error signal. 4. The model based process controller of claim 1, wherein the process model comprises one or more linear step response models or one or more impulse response models, and wherein the control unit implements a dynamic matrix control algorithm to produce the preliminary control signal. 5. The model based process controller of claim 1, wherein the process model is a first principles model. 6. The model based process controller of claim 1, wherein the process model is a first order plus deadtime process model. 7. The model based process controller of claim 1, wherein the process model is a second order process model. 8. The model based process controller of claim 7, wherein the process model is a second order plus deadtime process model. 9. The model based process controller of claim 1, wherein the control unit implements a model predictive control algorithm. 10. The model based process controller of claim 1, wherein the control unit implements a model predictive control algorithm including a Kalman filter. 11. The model based process controller of claim 10, wherein the Kalman filter is a general Kalman filter. 12. The model based process controller of claim 10, wherein the Kalman filter is a simplified Kalman filter. 13. The model based process controller of claim 1, herein the integrator is tunable. 14. The model based process controller of claim 13 wherein the integrator is tunable based on the factional deadtime of the process. 15. The model based process controller of claim 1, wherein the combiner unit comprises a summer. 16. The model based process controller of claim 1, wherein the process model unit uses one or more final control signal values as inputs to the process model to produce the predicted process variable value. 17. The model based process controller of claim 16, wherein the process model unit additionally uses one or more measured disturbance values within the process as inputs to the process model to produce the predicted process variable value. 18. A method of developing a set of process control signals for use in controlling a process, comprising: producing a set of predicted process variable values from a process model that models the operation of the process and from a set of controller output values provided as inputs to the process model;using the set of predicted process variable values to generate a set of preliminary control signals;developing an error signal from the set of predicted process variable values and a further set of process variable values;integrating the error signal to produce an integrated error signal; andcombining the integrated error signal with the set of preliminary control signals to produce the set of process control signals. 19. The method of developing a set of process control signals of claim 18, including using the set of process control signals to control the operation of the process. 20. The method of developing a set of process control signals of claim 18, wherein producing the set of predicted process variable values includes using a previous set of process control signals as the set of controller output values. 21. The method of developing a set of process control signals of claim 18, further including using one or more process measurements indicative of a controlled variable within the process along with the process model and the set of controller output values to produce the set of predicted process variable values. 22. The method of developing a set of process control signals of claim 18, further including using one or more process measurements indicative of measured disturbances within the process along with the process model and the set of controller output values to produce the set of predicted process variable values. 23. The method of developing a set of process control signals of claim 18, wherein developing the error signal from the set of predicted process variable values and a further set of process variable values includes determining an error between the set of predicted process variable values and a set of set points for a set of controlled variables. 24. The method of developing a set of process control signals of claim 18, wherein developing the error signal from the set of predicted process variable values and a further set of process variable values includes determining an error between the set of predicted process variable values and a set of measured values of a controlled variable. 25. The method of developing a set of process control signals of claim 18, wherein the process model comprises one or more linear step response models or one or more impulse response models, and wherein using the set of predicted process variable values to generate the set of preliminary control signals includes using a dynamic matrix control algorithm to produce the set of preliminary control signals. 26. The method of developing a set of process control signals of claim 18, wherein using the set of predicted process variable values to generate the set of preliminary control signals includes using a model predictive control algorithm to produce the set of preliminary control signals. 27. The method of developing a set of process control signals of claim 26, further including implementing a state observer in conjunction with model predictive control algorithm. 28. The method of developing a set of process control signals of claim 27, wherein the state observer is one of a general Kalman filter or a simplified Kalman filter. 29. The method of developing a set of process control signals of claim 18, further including enabling the integrator to be tuned. 30. The method of developing a set of process control signals of claim 18, further including enabling tuning of the integrator based on a factional deadtime of the process. 31. A process controller for use in controlling a process, comprising: a processor;a computer readable memory;a process model stored on the computer readable memory that models the operation of the process;a prediction routine stored on the computer readable memory that, when executed on the processor, uses a process control output signal and the process model to produce a predicted process variable value;a control routine stored on the computer readable memory that, when executed on the processor, uses the predicted process variable value and a process variable set point to generate a preliminary control signal;an error detection routine stored on the computer readable memory that, when executed on the processor, determines an error signal indicative of a difference between the predicted process variable value and a further process variable value;a tunable integrator routine stored on the computer readable memory that, when executed on the processor, integrates the error signal to produce an integrated error signal; anda combiner routine stored on the computer readable memory that, when executed on the processor, combines the integrated error signal with the preliminary control signal to produce a final control signal for use in controlling a manipulated process variable of the process. 32. The process controller of claim 31, wherein the prediction routine uses one or more process measurements indicative of a controlled variable within the process along with the process model and the process control output signal to produce the predicted process variable values. 33. The process controller of claim 31, wherein the prediction routine uses a previously calculated final control signal as the process control output signal. 34. The process controller of claim 33, wherein the further process variable value is a set point for a controlled variable and wherein the error detection routine develops the error signal as a difference between the predicted process variable value and the set point for the controlled variable. 35. The process controller of claim 33, wherein the further process variable value is a measured process variable value for a controlled variable and wherein the error detection routine develops the error signal as a difference between a previously predicted process variable value and the measured process variable value for the controlled variable. 36. The process controller of claim 33, wherein control routine implements a dynamic matrix control algorithm to generate the preliminary control signal. 37. The process controller of claim 33, wherein control routine implements a model predictive control algorithm to generate the preliminary control signal. 38. The process controller of claim 33, wherein control routine implements a model predictive control algorithm and a state observer algorithm to generate the preliminary control signal. 39. The process controller of claim 38, wherein the state observer algorithm comprises one of a general Kalman filter or a simplified Kalman filter. 40. The process controller of claim 31, wherein the tunable integrator is tunable to different integration settings based on a factional deadtime of the process.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.