Adaptive Fault Estimation for Dynamics of High Speed Train Based on Robust UKF Algorithm

TP13; This paper proposes an adaptive unscented Kalman filter algorithm ( ARUKF) to implement fault estimation for the dynamics of high?speed train ( HST) with measurement uncertainty and time?varying noise with unknown statistics. Firstly, regarding the actuator and sensor fault as the auxiliary variables of the dynamics of HST, an augmented system is established, and the fault estimation problem for dynamics of HST is formulated as the state estimation of the augmented system. Then, considering the measurement uncertainties, a robust lower bound is proposed to modify the update of the UKF to decrease the influence of measurement uncertainty on the filtering accuracy. Further, considering the unknown time?varying noise of the dynamics of HST, an adaptive UKF algorithm based on moving window is proposed to estimate the time?varying noise so that accurate concurrent actuator and sensor fault estimations of dynamics of HST is implemented. Finally, a five-car model of HST is given to show the effectiveness of this method.