A foreseeable Lyapunov function approach for H∞ asynchronous filtering of discrete-time switched systems

摘要

This paper investigates the H∞ asynchronous filtering problem of discrete-time linear switched systems. A foreseeable Lyapunov function is firstly proposed for the asynchronous filtering error system with mode-dependent integrated dwell time (MDIDT) switching. The novelty consists in that the Lyapunov function carries the future state in the form of the convex linear functions which enables us to take into account the state and noise information at the next instant during the design process of the filter. The Lyapunov function follows the operation of the asynchronous filter and is designed corresponding to the partition of the synchronous intervals. In light of these merits, the stability and non-weighted l2-gain performance can be deduced, and then the foreseeable H∞ asynchronous filter is devised to ensure the resulting filtering error system is globally uniformly exponentially stable (GUES) with a prescribed non-weighted l2-gain. The word foreseeable signifies that the H∞ performance analysis of the filtering error system can’t do without the noise knowledge at the next instant. Compared with the traditional filter design method, our method gets tighter bounds on the dwell time and smaller filtering error results. Finally, the effectiveness and superiority of our filtering scheme is verified by application examples.