An Improved UKF and Its Application in Maneuvering Target Tracking
Zehao Ye, Kai Yan, Hao Tu, Wei Han, Yuwen Luo, Yawei Song · 2023
UKF (unscented Kalman filter) is a widely used nonlinear Kalman filtering algorithm, but it needs accurate system model and noise to give good filtering results. Aiming at this problem, the paper puts forward an improved UKF filtering algorithm (IMP-UKF). Firstly based on the traditional UKF, the system state noise adaptive estimation equation of the SAGE-HUSA algorithm is introduced. Secondly, memory exponential decay weighting and covariance matching criterion are introduced into the system state noise adaptive estimation equation. Finally, tracking simulations are performed for maneuvering targets. The simulation results show that the IMP-UKF algorithm is able to adaptively adjust the state noise and its covariance matrix to fit the motion state of the target better under the conditions of inaccurate system state model and state noise. And it is almost not affected by the sudden change of the target maneuver. It maintained a good track of maneuvering target.