Dynamic Event-Triggered Intermittent Fault Detection for Time-Varying Stochastic Systems

In this article, the problem of intermittent fault (IF) detection is investigated for stochastic linear time-varying (LTV) systems using dynamic event-triggered methods. Using the nonuniform sampling approach, the event-triggered system is transformed into a time-varying system with varying sampling periods. Using the moving horizon estimation strategy, a new IF detection filter is designed to generate residual signals, which can be decoupled from event-triggered transmission errors and estimation errors. Moreover, an event-triggered IF detection algorithm is proposed such that the appearance time and disappearance time of IFs can be detected quickly for stochastic LTV systems. In order to analyze the detectability of IFs for systems with/without event-triggered cases, the concept of distinguishability is introduced for IFs. Sufficient conditions are derived to guarantee the detectability of IFs for LTV systems. Finally, an experiment concerning the rotary steerable drilling tool system is provided to illustrate the effectiveness of the proposed IF detection method.