非线性系统状态突变下的非退化粒子滤波方法研究

摘要

For the degeneracy and sample impoverishment caused by re-sampling in the particle filter, an improved particle filter algorithm based on particle swarm optimization was proposed by integrating the idea of particle swarm optimization into the sample stage of the particle filter. This work and its features were embodied mainly in the following two interrelated aspects: On the one hand, the algorithm selected the particle state with the maximum weight of the last moment as the optimal value before sampling, and then moved the particle re-sampled at the last moment to the high likelihood region around the optimal value according to the designed strategy, which could increase the diversity and effectiveness of the particles and avoid the degeneracy of particles effectively. On the other hand, the proposal distribution was proposed, and the computability of the new proposal distribution was proved theoretically. The simulation results show that the proposed algorithm performs better than UPF algorithm in accuracy and time, and has a strong adaptability to the mutation of nonlinear system state.%针对粒子滤波算法中的粒子退化及重采样所引起的粒子多样性减弱问题,将粒子群优化思想融合到粒子滤波的采样阶段,提出了一种改进的基于粒子群优化的粒子滤波算法.本项工作的特色主要表现在如下相互联系的两个方面:第一,在采样前,首先取上一时刻重采样前权重最大的粒子状态作为最优值,然后根据改进算法的粒子移动策略,将上一时刻重采样后的粒子移向最优值周围的高似然区域,从而能够增加粒子的多样性和有效性,有效避免了粒子的退化；第二,构造了改进算法的建议分布,并从理论的角度证明了该建议分布的可计算性.实验结果表明,从精度和时间这两个方面的综合考虑,改进算法要优于UPF等算法,对非线性系统突变具有更强的适应性.