Nonlinear MPC based on the state-space form of RBF-ARX model

Author

Hui Peng ; Shioya, Hideo ; Xiaoyan Peng ; Sato, Keisuke

Author_Institution

Sch. of Inf. Sci. & Eng., Central South Univ., Changsha, China

Volume

2

fYear

2004

fDate

2-4 Sept. 2004

Firstpage

1679

Abstract

A state-space form of the RBF-ARX model based nonlinear predictive controller is presented in this paper. The state-space representation proposed is built from an offline identified RBF-ARX model, and its state variables have no need to be estimated online using a state observer. The system nonlinearity is handled by applying the operating-point dependent RBF-ARX model, that is, a hybrid pseudo-linear time-varying model which is composed of Gaussian radial basis function (RBF) neural networks and linear ARX model structure. The nonlinear MPC proposed in this paper can make use of the future operating-point-related information, if that can be obtained in some cases, to improve control performance. Stability issue is discussed, and the effectiveness of the predictive control strategy presented is also showed through simulation result.

Keywords

Gaussian processes; control nonlinearities; neural nets; nonlinear control systems; observers; predictive control; radial basis function networks; state-space methods; time-varying systems; Gaussian radial basis function; hybrid pseudolinear time-varying model; model predictive control; neural networks; nonlinear control; state observer; state-space methods; system nonlinearity; Automotive engineering; Information science; Neural networks; Nonlinear dynamical systems; Observers; Predictive control; Predictive models; Radial basis function networks; Stability; State estimation;

fLanguage

English

Publisher

ieee

Conference_Titel

Control Applications, 2004. Proceedings of the 2004 IEEE International Conference on

Print_ISBN

0-7803-8633-7

Type

conf

DOI

10.1109/CCA.2004.1387618

Filename

1387618