Simultaneous Recursive Identification of Parameters and Switching Manifolds Identification of Discrete-Time Switched Linear Systems
Zengjie Zhang,Yingwei Du,Tong Liu,Fangzhou Liu,Martin Buss
DOI: https://doi.org/10.48550/arXiv.2303.04015
2023-03-08
Abstract:A novel procedure for the online identification of a class of discrete-time switched linear systems, which simultaneously estimates the parameters and switching manifolds of the systems, is proposed in this paper. Firstly, to estimate the parameters of the subsystems, a discrete-time concurrent learning-based recursive parameter estimator is designed to guarantee the exponential convergence of the estimation errors to zero. Secondly, as an assistant procedure of the identification framework, an online switching detection method is proposed by making use of the history stacks produced by the concurrent learning estimators. Thirdly, techniques of incremental support vector machine are applied to develop the recursive algorithm to estimate the system switching manifolds, with its stability proven by a Lynapunov-based method. At the end of the paper, the stability and precision of the proposed identification methods are confirmed by the numerical simulation of a 2-order switched linear system. Compared to the traditional offline identification methods, the proposed online identification framework possesses superior efficiency with respect to large amounts of data, while the limitations and outlook of this framework are also discussed within the conclusion.
Systems and Control