Fault Tolerant Attitude Estimation for an LEO Satellite Using a Multi-hypothesis Filter

Abstract

The aim of this paper is to improve state estimation and FDI performance. We proposed an algorithm using a model based multi-hypothesis filter (MHF). With this method, we can detect and identify the fault on a satellite actuator. Also, we can improve the attitude estimation performance of a satellite attitude control system. In this method, there are several models for a possible failure situation and the model probabilities are decided from the residuals of each filter. The master filter finds the most suitable state estimation result from the weighting sum of each filter based on probabilities. And a probability crossing detection algorithm and thresholding algorithm based on probabilities are introduced for fault detection and identification of satellite actuators. The simulation results demonstrate the advantages of the MHF-based method proposed.