On designing event-based H∞ filters for sampled-data systems

This paper is concerned with the H∞ filtering for sampled-data systems. First, an event-based data packet processor is introduced to release the sampled measurement outputs only if an event condition is violated. As a result, the communication resources can be significantly saved while the desired system performance can be preserved. Second, the resulting filtering error system is modeled as an interval time delay system. By employing Lyapunov-Krasovskii functional method, a new bounded real lemma (BRL) is formulated such that the filtering error system can achieve a prescribed H∞ performance level. Third, by performing an invertible linear transformation on the filtering error system, the corresponding BRL to the transformed filtering error system is obtained, from which, suitable H∞ filters and the threshold parameter in the event condition can be co-designed provided that a set of linear matrix inequalities are feasible. Finally, the effectiveness of the proposed method is demonstrated through a mechanical system with two masses and two springs.