摘要
针对粒子滤波中粒子退化和重采样造成的粒子多样性损失的问题,提出一种改进的粒子滤波算法,将蝠鲼觅食优化算法应用到粒子滤波当中。通过链式和螺旋阶段使驱使粒子向高似然区域靠近,粒子随机进行横向交叉,提高多样性;利用适应度阈值将粒子分为高低权值两组,分别进行翻滚觅食过程,通过全局最优值和低权值粒子进行线性组合,使后者仍以新粒子的形式存在于滤波之中,改善粒子退化现象。仿真结果表明,对比其它智能优化的粒子滤波算法,提出算法提高了最优解求解精度,减少了滤波的估计误差。
Aiming at the problems of particle degradation and diversity loss caused by resampling in particle filtering,an improved particle filtering algorithm was proposed,in which the manta ray foraging optimization algorithm was applied to particle filtering.Through the chain and spiral stages,the driven particles approached to the high likelihood region,and crossed the particles randomly to improve the diversity of particles.The particles were divided into two groups with high and low weights by fitness threshold,and the rolling foraging process was carried out respectively.The linear combination of the global optimal value and the low weight particles was carried out,so that the latter could still exist in the filtering as new particles form to improve the particle degradation.Simulation results show that compared with other intelligent optimization particle filtering algorithms,the proposed algorithm improves the solving accuracy of optimal solution,reduces the estimation error of filtering and improves the filtering accuracy.
作者
陆星辰
静大海
杨佳林
李文栋
黎瑶
LU Xing-chen;JING Da-hai;YANG Jia-lin;LI Wen-dong;LI Yao(Array and Information Processing Laboratory,College of Computer and Information,Hohai University,Nanjing 211100,China)
出处
《计算机工程与设计》
北大核心
2023年第9期2643-2649,共7页
Computer Engineering and Design
基金
国家重点研发计划基金项目(2018YFB1308700)。
关键词
粒子滤波
蝠鲼觅食优化
状态估计
粒子退化
粒子多样性
横向交叉
粒子分组
particle filter
manta ray foraging optimization
state estimation
particle degradation
particle diversity
horizontal cross
particle group
