An Improved Unscented Kalman Filter Algorithm for Maneuvering Target Tracking
Download as PDF
Jian Liu, Tian Yu
In order to improve the performance of the strong unscented Kalman filter (STUKF) for maneuvering target tracking and shorten the time required by the algorithm, a fast multi-fading STUKF algorithm with time-varying noise estimator is proposed. Considering the strategy of fading factor, the introduction position and reference method of the fading factor are improved. The filter gain and covariance matrix are adjusted adaptively. The process noise and observed noise with unknown statistical properties are estimated by adding a time-varying noise estimator to emphasize the importance of the adjacent data. The simulation results of various algorithms for the maneuvering target motion model are compared. The results show that the new filtering algorithm can track the maneuvering target better, improve the operating efficiency of UKF and the tracking accuracy. In the case of inaccurate noise statistics, the problem of filtering accuracy divergence is overcome, excessive calculation amount is avoided, and more efficient filtering performance is obtained.
Target tracking, Unscented kalman filter, Fading factor, Parameter adaptation, Noise estimator