摘要
针对粒子滤波算法重采样导致的样本贫化问题,提出基于鸽群优化(PIO)思想改进的粒子滤波算法。将鸽群不断从较远位置飞向适应度值高的地方的归巢过程引入到粒子滤波中,驱使粒子不断向高似然区域移动,并将自适应交叉操作加入到鸽群优化过程当中,保障样本多样性。通过非线性模型仿真实验表明,所提算法相对于标准粒子滤波算法,精度提高了45%,稳定性提高了72%,同时降低了状态估计所需的粒子数量。
A particle filtering（PF） algorithm based on pigeon-inspired optimization（PIO） is proposed,aiming at problem of sample impoverishment caused by resampling of PF algorithm.When pigeons fly,they usually fly from a far position to high fitness areas constantly.This optimum process is introduced into PF to drive particles moving towards high likelihood areas ceaselessly. And adaptive crossover operation is added to the process of PIO to guarantee the diversity of samples. The nonlinear model simulations experiments show that the precision of the proposed algorithm is improved by 45 % compared with the standard PF algorithm,the stability is improved by72 %,and the number of particles required for the state estimation is reduced at the same time.
作者
韩锟
张赫
HAN Kun;ZHANG He(School of Traffic ＆ Transportation Engineering,Central South University,Changsha 410075,China)
出处
《传感器与微系统》
CSCD
2018年第11期139-141,144,共4页
Transducer and Microsystem Technologies
基金
湖南省自然科学基金资助项目(2016JJ4117)
中南大学中央高校基本科研业务费专项资金资助项目(2017ZZTS810)
关键词
粒子滤波
样本贫化
鸽群优化算法
自适应交叉
状态估计
particle filtering （ PF ）
sample impoverishment
pigeon-inspired optimization （ PIO ） algorithm
adaptive crossover
state estimation
