Using semi-blind source separation in multi-sensor fault location of satellite attitude determination system

Abstract

Fault location is important for attitude determination (AD) of satellite which provides better services without any degradation. Firstly, the assumption of statistically independent between 3-axis Euler angles and statistically independent between Euler angles and faults are proposed in this paper. Also it is assumed the fault appears only in one group of sensors and the fault is additive. With two difference groups of sensor measuring the Euler angles of satellite, the link between fault location and blind source separation (BSS) is established. Secondly, independent component analysis(ICA) algorithm is used for BSS. Finally, using semi-blind source separation (SBSS) method with correlation coefficient and pattern recognition of the mixed matrix, ambiguities of BSS are overcome. The numerical simulation shows this method can fulfill fault location and extraction of the attitude AD system.