Distributed filtering method for fault diagnosis in a sensor network

摘要

Discussed herein are a sensor-network system and a method thereof for detecting and isolating faults occurring in sub-systems and sensors included in the network. The sub-systems and the sensor measurements are subject to noise and disturbances. A bank of Kalman-filters (KF) is used to detect and isolate the faults. Each KF is driven by either a pair of consecutive sensor measurements or a pair including a reference input and a sensor measurement. The KF residual is computed for each measurement pair and is a reliable indicator of a fault in subsystems and sensors located in the network.