一种比例尺蒙特卡罗滤波算法  

A Scale Monte Carlo Filtering Algorithm

作　　者：赵新[1] 宁小磊[1] 梁兴[1] 

机构地区：[1]中国华阴兵器试验中心,陕西华阴714200

出　　处：《弹箭与制导学报》2011年第6期179-182,共4页Journal of Projectiles,Rockets,Missiles and Guidance

基　　金：装备预研项目资助

摘　　要：粒子滤波是一种极具潜力的非线性滤波算法,粒子退化是粒子滤波的主要问题。为此,提出一种比例尺粒子滤波算法(SPF)。在重要性抽样之后,按粒子重要性权值大小,将其分为好粒子群和较差粒子群两个种群,使用固定的或动态变化的比例尺系数对好粒子群和较差粒子群中的元素求取加权值,从而生成一些备选粒子。并通过重要性权值实现粒子的优选,从而使最好粒子参与状态估计。仿真结果表明,该算法是有效的。The particle filtering is a promising non-linear filtering,but degeneracy phenomenon is a problem.In this paper,a new particle filter（scale particle filter,SPF） was proposed to overcome the above-mentioned problem.After important sampling step,the particles were classified into two groups according to their particle-weights,and then some candidate particles were generated by the scale mean of the two groups of particles.The optimizing selection of particles was realized based on its own particle-weight so that the better particles were obtained for estimation.Simulation results demonstrate the feasibility of SPF.

关 键 词：非线性滤波 粒子滤波 比例尺 粒子退化 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]