基于改进差分进化的高精度粒子滤波算法

针对智能优化粒子滤波算法精度较低和收敛速度慢的问题,提出一种改进适应度函数和搜索策略的差分进化粒子滤波算法(IDE-PF).该算法通过自适应融合粒子权值和量测误差得到适应度函数,并利用该函数评价粒子的可信度,引导粒子向后验概率密度取值高的位置移动,同时引入新的搜索策略,不仅保持了粒子多样性,还加快了算法收敛的速度.仿真结果表明,该算法可有效提高智能优化粒子滤波对于非线性系统状态估计的精度和实时性.

A High Precision Particle Filter Based on Improved Differential Evolution

Abstract： A particle filter based on differential evolution with improved fitness function and search strategy was proposed to solve the problem of the low precision and slow convergence rate of particle filters based on intelligent optimization algorithms. The algorithm defined a new fitness function based on adaptive fusion of particle weight and its measurement error. The function was used to evaluate the credibility of the particles and move them to positions with larger values of posterior density function. Synchronously, a new search strategy was introduced to differential evolution which could maintain the diversity of the particles and accelerate the convergence rate of the particle filter. Experiment results show that the proposed algorithm effectively improves the accuracy and real time performance of the intelligent optimization particle filter for nonlinear system state estimation.