Event-Triggered Fault Detection and Diagnosis for Networked Systems with Sensor and Actuator Faults

This paper investigates the problem of fault detection, isolation, and estimation for a networked system with actuator and sensor faults. To deal with the bandwidth constraint, an event-triggered scheduling mechanism is utilized to determine whether the sensor observation shall be transmitted to the fault filter according to the importance of information. In this study, two independent Markovian jump chains are introduced to describe the temporal occurrence of sensor fault and the random switching between the normal condition and the faulty ones of the actuator, respectively. To alleviate the compromise between the model number of fault models and computational complexity in the existing interacting multiple models (IMM) approaches, a novel event-triggered fault detection and diagnosis algorithm is proposed based on the augmented IMM framework, where the fault location to be detected is added into the model set and the fault amplitude to be estimated is augmented into the system state. Finally, a Monte Carlo simulation involving tracking a two-dimension moving target is provided to illustrate the effectiveness and efficiency of the proposed method.