The problems of making the Kalman filter robust for contaminated observation are considered in this paper. Two typical robust estimation algorithms are investigated for the suppression of outliers which appeared in observation of Kalman filter. A new approach focus on how to distinguish and eliminate the outliers base on the variational rate of observation which sampled from continuous two nodes of certain sensor is proposed. The eminent character is that it can guarantee the estimation without deviation regardless of the outliers's distributing. The feasibility of the approach is demonstrated by the simulation.
展开▼