基于粒子筛选处理的粒子滤波改进算法

摘要

针对粒子滤波中的粒子贫化问题,分析了目前用于增加粒子多样性方法存在的不足,提出了一种新的粒子筛选与处理方法.通过设置筛选区间,保留该区间内的粒子,对区间外的粒子进行移动处理,从而改善粒子分布.仿真结果表明,该方法能够有效缓解粒子贫化问题,提高滤波精度.同时由于有效样本数增加,降低了重采样次数,总体上减少了算法运行时间.%To solve the particle impoverishment problem in particle filter (PF),an improved algorithm is proposed based on a new particle selection and process strategy.By choseing filtering interval,the predicted particles falling into the desired interval are accepted,and the others outside the interval are corrected by concentrating them from remote areas to high likelihood areas of probability density function.Experimental results show that the new algorithm increases the filtering accuracy,compared with standard PF.Meanwhile,as the number of effective particle increases,the requirement of resampling operation is decreased,thus reducing the computational time.