IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0002158
(2004-12-02)
|
등록번호 |
US-7317953
(2008-01-08)
|
발명자
/ 주소 |
- Wojsznis,Wilhelm K.
- Blevins,Terrence L.
- Nixon,Mark J.
- Wojsznis,Peter
|
출원인 / 주소 |
- Fisher Rosemount Systems, Inc.
|
대리인 / 주소 |
Marshall, Gerstein & Borun LLP
|
인용정보 |
피인용 횟수 :
14 인용 특허 :
26 |
초록
▼
An adaptive multivariable process control system includes a multivariable process controller, such as a model predictive controller, having a multivariable process model characterized as a set of two or more single-input, single-output (SISO) models and an adaptation system which adapts the multivar
An adaptive multivariable process control system includes a multivariable process controller, such as a model predictive controller, having a multivariable process model characterized as a set of two or more single-input, single-output (SISO) models and an adaptation system which adapts the multivariable process model. The adaptation system detects changes in process inputs sufficient to start an adaptation cycle and, when such changes are detected, collects process input and output data needed to perform model adaptation. The adaptation system next determines a subset of the SISO models within the multivariable process model which are to be adapted, based on, for example, a determination of which process inputs are most correlated with the error between the actual (measured) process output and the process output developed by the multivariable process model. The adaptation system then performs standard or known model switching and parameter interpolation techniques to adapt each of the selected SISO models. After the adaptation of one or more of the SISO models, the resulting multivariable process model is validated by determining if the adapted multivariable process model has lower modeling error than the current multivariable process model. If so, the adapted multivariable process model is used in the multivariable controller.
대표청구항
▼
What is claimed is: 1. A method of adapting a multivariable process model made up of two or more single-input single-output (SISO) models for use in a process controller that uses the multivariable process model to perform process control, the method comprising: selecting a subset of the SISO model
What is claimed is: 1. A method of adapting a multivariable process model made up of two or more single-input single-output (SISO) models for use in a process controller that uses the multivariable process model to perform process control, the method comprising: selecting a subset of the SISO models to adapt; individually adapting each of the selected subset of the SISO models; and providing the adapted subset of the SISO models to the process controller to be used in the multivariable process model. 2. The method of adapting a multivariable process model of claim 1, further including determining when to perform a model adaptation based on a change to a process input or a process output. 3. The method of adapting a multivariable process model of claim 2, wherein determining when to perform a model adaptation includes collecting and storing data indicative the process input or the process output and analyzing the collected data to detect a change in the process input or the process output greater than a predetermined amount. 4. The method of adapting a multivariable process model of claim 1, wherein selecting a subset of the SISO models includes analyzing one or more of the SISO models associated with an output of the process to determine a correlation measure between each of the one or more of the SISO models and the output of the process and selecting a SISO model to adapt based on the correlation measures. 5. The method of adapting a multivariable process model of claim 1, wherein selecting a subset of the SISO models includes determining, for each of a number of process inputs, a correlation measurement between the process input and an error measurement between a measured process output and a process output developed by the multivariable process model, using the correlation measurements to select one of the process inputs, and selecting one of the SISO models which relates the selected one of the process inputs to the measured process output as one of the subset of the SISO models. 6. The method of adapting a multivariable process model of claim 5, wherein using the correlation measurements to select one of the process inputs includes determining if a first one of the process inputs experienced a predetermined amount of change and selecting the first one of the process inputs as the selected one of the process inputs only if the first one of the process inputs experienced a predetermined amount of change. 7. The method of adapting a multivariable process model of claim 1, wherein selecting a subset of the SISO models includes determining which process inputs are most correlated with an error between a measured process output and a modeled process output developed by the multivariable process model and selecting one or more of the SISO models based on the correlation determination. 8. The method of adapting a multivariable process model of claim 1, wherein separately adapting each of the selected subset of the SISO models includes performing a model switching adaptation technique on at least one of selected subset of the SISO models. 9. The method of adapting a multivariable process model of claim 1, wherein separately adapting each of the selected subset of the SISO models includes performing an attribute interpolation adaptation technique on at least one of selected subset of the SISO models. 10. The method of adapting a multivariable process model of claim 1, further including validating the multivariable process model with the adapted subset of the SISO models prior to providing the adapted subset of the SISO models to the process controller to be used in the multivariable process model. 11. The method of adapting a multivariable process model of claim 10, wherein validating the multivariable process model with the adapted subset of the SISO models includes determining if the multivariable process model with the adapted subset of the SISO models has a lower modeling error than the multivariable process model without the adapted subset of the SISO models. 12. The method of adapting a multivariable process model of claim 10, wherein validating the multivariable process model with the adapted subset of the SISO models includes determining if the multivariable process model with the adapted subset of the SISO models operates on the same process input and process output data to provide better control than the multivariable process model without the adapted subset of the SISO models. 13. The method of adapting a multivariable process model of claim 1, further including transforming at least one of the adapted subset of the SISO models into a form used by the process controller prior to providing the adapted subset of the SISO models to the process controller to be used in the multivariable process model. 14. The method of adapting a multivariable process model of claim 1, wherein individually adapting each of the selected subset of the SISO models includes adapting a parameter based SISO model. 15. The method of adapting a multivariable process model of claim 14, wherein the parameter based SISO model is a first-order-plus-dead-time model. 16. The method of adapting a multivariable process model of claim 1, wherein individually adapting each of the selected subset of the SISO models includes adapting a non-parametric based SISO model. 17. The method of adapting a multivariable process model of claim 16, wherein the non-parametric based SISO model is one of a step response model and an impulse response model. 18. The method of adapting a multivariable process model of claim 16, wherein individually adapting each of the selected subset of the SISO models includes adjusting an attribute of the non-parametric based SISO model. 19. The method of adapting a multivariable process model of claim 18, wherein adjusting an attribute of the non-parametric based SISO model includes adjusting one of the time until a response is first observed, a scaling, and a slope of the non-parametric based SISO model. 20. A process control system for use in controlling a process, comprising: a process controller including a multivariable process model made up of two or more single-input single-output (SISO) models for use in controlling the process; and a model adaptation unit communicatively connected to the process controller, including; a first unit for selecting a subset of the SISO models for adaptation; a second unit for altering each of the selected subset of the SISO models; and a third unit for providing the altered subset of the SISO models to the process controller to be used in the multivariable process model. 21. The process control system of claim 20, wherein the model adaptation unit further includes a supervisor unit for detecting changes in one or more process inputs or process outputs to determine when to start an adaptation cycle. 22. The process control system of claim 21, wherein the supervisor unit includes a data collection unit, collects and stores data indicative of one of the one or more process inputs or process outputs in the data collection unit and analyzes the collected data to detect a change in the one of the one or more process inputs or process outputs greater than a predetermined amount. 23. The process control system of claim 21, wherein the first unit selects a subset of the SISO models by analyzing one or more of the SISO models associated with an output of the process to determine a correlation measure between an input of each of the one or more of the SISO models associated with the output of the process and the output of the process, and selects a SISO model to adapt based on the correlation measures. 24. The process control system of claim 20, wherein the first unit determines, for each of a number of process inputs, a correlation measurement between the process input and an error measurement between a measured process output and a process output developed by the multivariable process model, uses the correlation measurements to select one of the process inputs, and selects one of the SISO models which relates the selected one of the process inputs to the measured process output as one of the subset of the SISO models. 25. The process control system of claim 20, wherein the second unit performs a model switching model adaptation technique on at least one of selected subset of the SISO models. 26. The process control system of claim 20, wherein the second unit performs an attribute interpolation model adaptation technique on at least one of selected subset of the SISO models. 27. The process control system of claim 20, wherein the model adaptation unit further includes a validation unit adapted to test the multivariable process model using the altered subset of the SISO models prior to the third unit providing the altered subset of the SISO models to the process controller to be used in the multivariable process model. 28. The process control system of claim 27, wherein the model adaptation unit further includes a transformation unit for transforming at least one of the altered subset of the SISO models into a form used by the multivariable process model within the process controller prior to providing the altered subset of the SISO models to the process controller to be used in the multivariable process model. 29. The process control system of claim 20, wherein at least one of the selected subset of the SISO models is a parameter based SISO model. 30. The process control system of claim 29, wherein the parameter based SISO model is a first-order-plus-dead-time model. 31. The process control system of claim 20, wherein at least one of the selected subset of the SISO models is a non-parametric based SISO model. 32. The process control system of claim 31, wherein the non-parametric based SISO model is one of a step response model and an impulse response model. 33. The process control system of claim 31, wherein the second unit adjusts an attribute of the non-parametric based SISO model by adjusting one of the time until a response is first observed, a scaling, and a slope of the non-parametric based SISO model. 34. The process control system of claim 20, wherein the process controller is a model predictive control type controller. 35. A model adaptation unit for use in a process control system having a multivariable process model made up of two or more single-input single-output (SISO) models, the model adaptation unit comprising: a computer readable medium; a program stored on the computer readable medium and implemented on a processor, the program including; a first routine configured to select a subset of the SISO models for adaptation; a second routine configured to adapt each of the selected subset of the SISO models; and a third routine configured to provide the adapted subset of the SISO models to the multivariable process model. 36. The model adaptation unit of claim 35, wherein the program further includes a supervisor routine configured to detect a change in a process input or process output to determine when to start an adaptation cycle. 37. The model adaptation unit of claim 35, wherein the first routine determines, for each of a number of process inputs, a correlation measurement between the process input and an error measurement between a measured process output and a process output developed by the multivariable process model, uses the correlation measurements to select one of the process inputs, and selects one of the SISO models which relates the selected one of the process inputs to the measured process output as one of the subset of the SISO models. 38. The model adaptation unit of claim 35, wherein the second routine performs a model switching model adaptation technique on at least one of selected subset of the SISO models. 39. The model adaptation unit of claim 35, wherein the second routine performs an attribute interpolation model adaptation technique on at least one of selected subset of the SISO models. 40. The model adaptation unit of claim 35, wherein the program further includes a validation routine that tests the multivariable process model using the adapted subset of the SISO models prior to the third routine providing the adapted subset of the SISO models to the multivariable process model. 41. The model adaptation unit of claim 40, wherein the program further includes a transformation unit that transforms at least one of the adapted subset of the SISO models into a form used by the multivariable process model prior to the third routine providing the adapted subset of the SISO models to the multivariable process model. 42. The model adaptation unit of claim 35, wherein at least one of the selected subset of the SISO models is a non-parametric based SISO model and wherein the second unit adjusts an attribute of the non-parametric based SISO model by adjusting one of the time until a response is first observed, a scaling, and a slope of the non-parametric based SISO model.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.