Asymptotic fault-tolerant fusion algorithm based on P-value mapping for multi-sensors navigation

Abstract

The fault tolerance capacity of fusion algorithm is critical for the robustness and stability of multi-sensors navigation system in failure. In traditional fault-tolerant fusion algorithm, the fault subsystem is isolated entirely on the fault- detected epoch. This strategy ignored the asymptotic changing and components difference of soft failure influence. In this paper, an asymptotic fault-tolerant fusion algorithm based on P-value mapping is proposed for multi-sensors navigation. The failure influence mapped on local estimation components is modelled. Then quality of local estimation influenced by soft failure is evaluated according to P-value theory. The estimated quality values of local estimations are used as reference for fault-tolerant tuning in global fusion. Simulations of INS/GPS/SAR integrated navigation using the traditional federal filter and the proposed fault-tolerant algorithm are carried out. The simulation results indicate that the fault-tolerant performance under soft failure is improved significantly by using proposed algorithm.