Research on particle filter algorithm for improving particle diversity
Qiduo Liu, Yongqiang Bai · 2017
The particle filter has been applied widely for non-linear filtering since it can relax the linear and Gaussian assumptions. However, the traditional particle filter will cause particle depletion, losing the particle diversity. Aiming at this problem, an improved particle filter based on adaptive chaos immune genetic resampling is proposed. The interval of chaotic traversal changes adaptively to further speed up the local search. A novel selection mechanism considering the antibody concentration and affinity is proposed to increase the population diversity by replicating particles with high weights and low concentration dynamically. Experimental results show that the improved method has higher filtering precision and faster running speed than conventional filtering algorithm and common resampling method based on chaos immune genetic algorithm, and the diversity of the population is improved remarkably.