基于混沌的萤火虫改进粒子滤波算法研究

针对常规的粒子滤波算法存在粒子权值退化和采样粒子贫化以及需要大量粒子才能进行比较准确的状态估计的问题,提出了一种基于混沌的萤火虫改进粒子滤波算法.利用混沌系统所具有的遍历性和随机性初始化粒子群,使得初始粒子分布更加均匀,同时向常规粒子滤波算法中引进萤火虫算法的寻优机制,使得粒子能够向高似然区域运动,提高了滤波精度,并对部分权值优秀粒子进行混沌细搜索,对部分权值低的粒子进行再生,提高了种群多样性.实验表明:该方法尤其是在粒子种群数量较小的情况下,较常规粒子滤波精度更高,并有效地改善了权值退化和样本贫化问题.

Research of improved particle filtering algorithm for fireflies based on chaos

Aiming at problem exists in conventional particle filtering algorithm of particle weight degradation and sampling particle impoverishment and it needs for a large number of particles to achieve accurate state estimation a firefly improved particle filtering algorithm based on chaotic is proposed.By using the ergodic and random initialization particle swarm of the chaotic system,the initial particle distribution is more uniform.At the same time,the optimization mechanism of the firefly algorithm is introduced into the conventional particle filtering algorithm,which can make the particles move to the high likelihood region,so as to improve the precision of the filtering.Using chaos search on the part of the weight of the excellent particle at the same time,and regenerate part of low weight particles,to achieve demand of diversity of improving population.Experiments show that this method is more accurate than the conventional particle filtering,especially in the case of small populations,and effectively improve the weight degeneracy and sample impoverishment problem.