A zonotope-based fault detection for multirate systems with improved dynamical scheduling protocols

Abstract

This paper investigates the distributed fault detection problem for a class of discrete-time multi-rate systems over sensor networks with an improved communication protocol, where both unknown-but-bounded (UBB) noises and faults are comprised in two zonotopic sequences. Inspired by famous try-once-discard (TOD) protocols, a novel quasi-TOD-based protocol with dynamical weights is proposed to govern the signal transmission between the sensor nodes and the remote filter. In terms of lifting techniques, the augmented filtering error dynamics is set up a unified sampling rate. By resorting to the properties of zonotopes as well as the mathematical induction, a series of zonotopic sets confining filtering errors is first received, and then their F-radii are minimized to derive the desired filter parameters. Furthermore, a zonotope-based residual evaluation mechanism is designed to detect possibly occurred faults. Finally, a numerical example is employed to verify the availability of the presented zonotope-based fault detection scheme.