Fault Diagnosis for Networked Systems By Asynchronous IMM Fusion Filtering

A fault diagnosis method is proposed based on networked asynchronous interacting multiple model (IMM) fusion filtering for a kind of networked systems with multiple asynchronous sensors and stochastic packet dropouts. The proposed networked asynchronous IMM fusion filtering algorithm is used to perform fusion filtering for the model set consisting of the normal model and all kinds of possible fault models of the system, where different arriving probabilities are considered for different sensor communication channels and the unknown fault amplitude is taken as the augmented system state. Fault detection and location are achieved based on the model possibilities, and the estimates of system state and fault amplitude can be obtained simultaneously. The proposed method avoids the problem of determining fault amplitude in the model set design of traditional IMM fault diagnosis approaches, improves the accuracy of fault diagnosis by fusing information from multiple sensors, and can be used to asynchronous multi-sensor networked systems with arbitrary sampling rates and arbitrary initial sampling time instants. The feasibility and effectiveness of the proposed algorithm are illustrated by simulation examples.