Redundant design can improve the performance of an inertial navigation system, and the configuration is an important factor in determining the accuracy and reliability of the redundant inertial navigation system. As well as the sensor quantity and the reliability of single sensors, the reliability of the redundant inertial navigation system also depends on the capability of the fault diagnosis, which has been neglected in much of the previous research. This paper verifies that configurations designed using the previous method cannot maintain optimal performance in fault diagnosis. To optimize the performance in fault detection and isolation through the configuration design, an evaluation function for fault detectability and isolability is proposed based on Kullback–Leibler divergence. On this basis, the inherent diagnosability analysis method is proposed, and the configuration design method is developed with the analysis results. Simulation experiments prove that the proposed diagnosability analysis method can find the defects in the fault detection and isolation, and the diagnosability can be assessed correctly. Configurations designed using the proposed methodensure the optimal accuracy and reliability of the system, and effectively improve the fault diagnosis ability, too.

No abstract