FDI using Multiple Parity Vectors for Redundant Inertial Sensors

This paper discusses fault detection, isolation, and accommodation (FDIA) problem of redundant sensors in inertial navigation systems (INS). The averaged parity vector (APV) method is suggested to improve the probability of correct isolation and reduce the numbers of false alarms and wrong isolation. The proposed method employs a new accommodation threshold based on the error covariance of an estimated variable, which is related to the navigation accuracy of INS. This accommodation threshold is also used as the fault detection threshold, so the APV algorithm detects only non-tolerable faults. Given a conditional probability of correct isolation, the minimum number of sample measurements is given with respect to fault size. Computer simulation and experimental results verify the validity of the proposed method.