摘要
为解决粒子滤波算法中粒子权值退化和粒子贫化问题,提出基于布谷鸟算法优化的粒子滤波(CS-PF)。对鸟窝寻优位置公式增加扰动项,将粒子集合平分为log2N (N为粒子数)项,降低粒子移动时陷入局部最优的可能性;为保证粒子多样性,当均匀分布的随机数大于阈值时,利用改进的鸟窝寻优方法更新粒子位置,部分粒子向高似然区移动,部分在低似然区。实验结果表明,粒子位置更新阈值在[0.2,0.35]时,CS-PF能够有效解决粒子滤波的权值退化和粒子贫化问题。
To solve the problem of particle weight degradation and particle dilution in particle filter algorithm,an improved particle filter algorithm based on cuckoo optimization algorithm(CS-PF)was proposed.Disturbance item was added to nest optimized formula,and the set of sampling particle was divided into log 2 N(N is the number of particle)terms to reduce the possibility that particles fell into local optima when moving.To ensure particle diversity,when the uniform distribution random number was bigger than threshold,improved nest optimized formula was used to update position of particles,parts of particles were then in high likelihood region,and others were in low likelihood region.Experimental results show that when the updating threshold of particle position is between 0.2 and 2.35,CS-PF can effectively solve the particle filter weight degeneracy and particle impoverishment problem.
作者
白晓波
邵景峰
和征
田建刚
BAI Xiao-bo;SHAO Jing-feng;HE Zheng;TIAN Jian-gang(School of Management,Xi’an Polytechnic University,Xi’an 710048,China;Department of Engineering Foundation,Army Academy of Border and Coastal Defence,Xi’an 710108,China)
出处
《计算机工程与设计》
北大核心
2018年第11期3401-3407,共7页
Computer Engineering and Design
基金
陕西省工业科技攻关基金项目(2017GY-039)
陕西省教育厅服务地方专项计划基金项目(16JF009)
2017年中国纺织工业联合会科技指导性计划基金项目(2017067)
2017年"纺织之光"中国纺织工业联合会高等教育教学改革基金项目(2017BKJGLX160)
关键词
权值退化
粒子贫化
布谷鸟算法
粒子滤波
数据拟合
weight degradation
particle dilution
cuckoo algorithm
particle filter
data fitting
作者简介
白晓波(1983),男,陕西汉中人,硕士,工程师,CCF会员,研究方向为智能信息处理、数据融合;邵景峰(1980),男,甘肃定西人,博士,副教授,研究方向为智能信息处理;和征(1978),男,陕西渭南人,博士,讲师,研究方向为制造业服务化、物流与供应链管理;田建刚(1982),男,陕西宝鸡人,讲师,研究方向为智能信息处理、数据融合。
