A Novel Scheme for Fault Detection Using Data-Driven Gap Metric Technique

Abstract

This paper considers the fault detection problem for uncertain linear time-invariant systems. Based on the data-driven computational method for the gap metric, a fault detection scheme is designed by monitoring the gap metric between the running process and its nominal system with the direct use of offline and online data. Moreover, an alternative iterative realization of the stable image representation is proposed, based on which the gap metric is obtained and the fault detection is conducted with less calculation efforts. In addition, owing to the physical properties behind the gap metric, reliability analysis for systems with multiplicative faults is addressed. The numerical simulation examples are presented to demonstrate the effectiveness of the fault detection scheme.