Parity space‐based optimal event‐triggered fault detection

Abstract

The article investigates the problem of parity space-based optimal fault detection for event-triggered systems. By introducing a novel parity vector based event-triggering mechanism, and taking the effects of faults, disturbances and event-triggered transmission errors on the residual into consideration, it is proven that the optimal parity vector for time-triggered residual generator, which can make a trade-off between fault sensitivity and disturbance robustness, is also optimal for event-triggered counterpart, as both fault and disturbance information are included in the event-triggered transmission error. Then, a time-varying threshold is designed for residual evaluation by accounting for the effects of disturbance and event-triggered transmission errors on the residual. Based on this, a quantitative analysis of the influence of the event-triggering mechanism on false alarm rate is developed, which provides a way to choose the event-triggered parameter from the viewpoint of fault diagnosis. Moreover, the proposed method is extended to event-triggered systems with packet dropouts. Finally, a benchmark model of the vehicle lateral dynamic system is utilized to demonstrate the effectiveness and superiority of the proposed parity space-based optimal event-triggered fault detection approach.