Robust extended Kalman filtering for nonlinear systems with multiplicative noises

In this paper, we investigate the robust filter design problem for nonlinear systems with multiplicative noises. The aim of the problem is to design a state estimator with a predictor–corrector structure, such that the upper bound on the state estimation error variance is minimized. A robust extended Kalman filter (REKF) is proposed based on a novel method to obtain the upper bound on the variances of the multiplicative noises. Further analysis shows that the proposed filter guarantees a bounded energy gain from the multiplicative noises to the estimation error. The REKF is implemented on the satellite attitude determination system that consists of the gyroscopes and the star sensors. Its performance is illustrated by using the real data obtained from a gyroscope. Simulation results show that the REKF outperforms another robust algorithm. Copyright © 2010 John Wiley & Sons, Ltd.