Real-time scheduling of linear models for control and estimation
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06G-007/48
G05B-017/02
출원번호
US-0473413
(2009-05-28)
등록번호
US-9618919
(2017-04-11)
발명자
/ 주소
Viassolo, Daniel Edgardo
Taware, Avinash Vinayak
Brunell, Brent Jerome
출원인 / 주소
GENERAL ELECTRIC COMPANY
대리인 / 주소
Darling, John P.
인용정보
피인용 횟수 :
0인용 특허 :
3
초록▼
A method of generating linear models for a physical system of interest is implemented in steps including, first, determining offline, a set of linear models for the physical system of interest by linearization of a nonlinear computational model of the physical system of interest at selected operatin
A method of generating linear models for a physical system of interest is implemented in steps including, first, determining offline, a set of linear models for the physical system of interest by linearization of a nonlinear computational model of the physical system of interest at selected operating points or from desired data; second, analyzing offline, accuracy of each linear model and eliminating inaccurate linear models therefrom to provide a residual set of linear models; third, generating offline, linear models corresponding to grid points of one or more lookup tables based on the residual set of linear models; fourth, associating offline, lookup table grid points with selected scheduling variables, and fifth, generating algorithmic software for the physical system therefrom such that linear models for the physical system generated offline form the basis for online scheduling of linear models.
대표청구항▼
1. A method of generating linear models for an aircraft engine system, the method comprising: determining, offline, a set of linear models for the aircraft engine system by linearization of a nonlinear model of the aircraft engine system at selected operating points or from desired data;analyzing, o
1. A method of generating linear models for an aircraft engine system, the method comprising: determining, offline, a set of linear models for the aircraft engine system by linearization of a nonlinear model of the aircraft engine system at selected operating points or from desired data;analyzing, offline, accuracy of each linear model and eliminating inaccurate linear models therefrom to provide a residual set of the linear engine models;generating, offline, linear models corresponding to grid points of one or more rectangular lookup tables based on the residual set of the linear engine models;associating, offline, lookup table grid points or the residual set of the linear engine models with selected scheduling variables; andgenerating, offline, algorithmic software for the aircraft engine system therefrom such that the linear models for the aircraft engine system generated offline form a basis for online scheduling of linear models. 2. The method according to claim 1, wherein the desired data are selected from a set of flight envelope and aircraft engine parameters. 3. The method according to claim 2, wherein the set of parameters is selected from engine speed, altitude, power level, ambient temperature. 4. The method according to claim 1, wherein analyzing, offline, accuracy of each linear model and eliminating inaccurate linear models therefrom to provide the residual set of linear models comprises preliminary linear model filtering by comparing, for a number of linear models, dynamic system responses of the number of linear models against similar dynamic system responses of nonlinear computational models. 5. The method according to claim 4, wherein the dynamic system responses comprise step responses. 6. The method according to claim 4, wherein the analyzing, offline, accuracy of each linear model and eliminating inaccurate linear models therefrom to provide the residual set of linear models further comprises generating system metrics for each linear model and eliminating linear models exceeding predefmed conditions for the system metrics that render those linear models as outliers with respect to the system metrics. 7. The method according to claim 6, wherein the predefmed conditions for the system metrics are selected from dynamic parameters and steady-state parameters. 8. The method according to claim 7, wherein the dynamic parameters are selected from natural frequency and damping. 9. The method according to claim 7, wherein the steady-state parameters are selected from steady-state gains from control inputs to controlled outputs, steady-state gains from disturbances inputs to controlled outputs, and steady-state gains from control inputs/disturbances to constrained outputs. 10. The method according to claim 6, wherein the predefined conditions are selected from a predetermined range of data points and predetermined data point sign conditions. 11. The method according to claim 1, wherein generating, offline, linear models corresponding to grid points of one or more lookup tables based on the residual set of linear models comprises generating the linear models using a polytopic approach using the residual set of linear models as vertex linear models. 12. The method according to claim 1, wherein the basis for online scheduling of linear models comprises interpolation using lookup tables for each element of the linear model. 13. The method according to claim 12, wherein the basis for online scheduling of linear models comprises interpolation using lookup tables for each entry of state-space matrices representing the linear model. 14. The method according to claim 12, wherein interpolation using lookup tables for each element of the linear model comprises generation of a set of linear model element lookup tables rectangular with respect to the scheduling variables if the residual set of linear models are not rectangular with respect to the scheduling variables. 15. The method according to claim 14, wherein generation of the set of linear model element lookup tables rectangular with respect to the scheduling variables is based on using linear interpolation of elements of the residual set of linear models. 16. The method according to claim 14, wherein generation of the set of linear model element lookup tables rectangular with respect to the scheduling variables is based on using a polytopic method operating on the residual set of linear models. 17. The method according to claim 1, wherein the basis for online scheduling of linear models comprises a polytopic method that uses offline grid point linear models to generate online scheduled linear models. 18. The method according to claim 1, further comprising validating the online scheduling of linear models. 19. The method according to claim 18, wherein validating the online scheduling of linear models comprises testing the online scheduling against test linear models not used as grid or vertex linear models. 20. The method according to claim 18, wherein validating the online scheduling method comprises selecting a set of scheduling variables to be used for interpolation. 21. The method according to claim 20, wherein selecting the set of scheduling variables comprises comparing predefined metrics of a test set of linear models to true values when different sets of scheduling variables are used for scheduling or interpolating the test linear models and selecting the set of scheduling variables that gives the closest match. 22. The method according to claim 1, wherein the online scheduling comprises generating current linear models online via the algorithmic software in response to the linear models for the aircraft engine system generated offline. 23. The method according to claim 1, wherein the online linear model scheduling comprises real-time implementation of online linear model scheduling comprising using a lookup table constructed offline to obtain indices of a predetermined number of grid or vertex linear models closest in a scheduling variable space selected for interpolation using an element by element table lookup or a polytopic approach. 24. The method according to claim 1, wherein the basis for online scheduling of linear models comprises interpolating using lookup tables for each element generated using a grid of linear models rendered rectangular with respect to a selected set of scheduling variables using element by element interpolation or a polytopic approach. 25. The method according to claim 1, wherein the basis for online scheduling of linear models comprises a polytopic method that uses offline grid point linear models to generate online scheduled linear models. 26. A method of generating linear engine models offline for use in subsequent online scheduling of current linear engine models, the method comprising : determining a primary set of linear engine models by linearization of a nonlinear computational model of an aricraft engine at selected operating points or from desired data;analyzing accuracy of each primary linear engine model and eliminating inaccurate primary linear engine models therefrom to provide a secondary set of linear engine models;generating the linear engine models corresponding to grid points of one or more lookup tables based on the secondary set of linear engine models;associating lookup table grid points with selected scheduling variables; andgenerating algorithmic software configured to assist online scheduling of linear engine models in response to the lookup table grid points and the selected scheduling variables. 27. The method according to claim 26, wherein the desired data is selected from physical and operating parameters of the aircraft engine. 28. The method according to claim 27, wherein the desired data comprises engine speed, altitude, engine power level, and ambient temperature. 29. The method according to claim 26, wherein analyzing the accuracy of each primary linear engine model and eliminating inaccurate primary linear engine models therefrom to provide a secondary set of linear engine models comprises preliminary linear engine model filtering by comparing, for a number of primary linear engine models, dynamic responses of primary linear engine models against dynamic responses of nonlinear engine computational models. 30. The method according to claim 29, analyzing the accuracy of each primary linear engine model and eliminating inaccurate primary linear engine models therefrom to provide a secondary set of linear engine models further comprises generating metrics for each primary linear engine model and eliminating primary linear engine models exceeding predefmed conditions. 31. The method according to claim 30, wherein the predefmed conditions are selected from dynamic parameters and steady-state parameters. 32. The method according to claim 31, wherein the dynamic parameters are selected from natural frequency and damping. 33. The method according to claim 31, wherein the steady-state parameters are selected from steady-state gains from control inputs to controlled outputs, steady-state gains from disturbances inputs to controlled outputs, and steady-state gains from control inputs/disturbances to constrained outputs.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (3)
Volponi,Allan J.; Brotherton,Thomas, Empirical tuning of an on board real-time gas turbine engine model.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.