Improved UKF Algorithm Based on Cauchy Robust Function

The statistics of noise are usually unknown in most actual systems. Moreover, it is known that the Unscented Kalman Filter (UKF) has poor robustness on the inaccurate noise information, which also makes the filter accuracy rapidly decrease, even diverge. By using the robust correction of data, this work proposes an improved UKF algorithm based on Cauchy Robust function. The noise estimation is adaptively corrected by the combining weighted function, which is calculated from the residual error between process measurement and its estimation. Thus, the accuracy of the UKF algorithm can be increased. Finally, two simulation examples are provided to show the effectiveness of the proposed algorithm.