基于对称KL距离的自适应粒子滤波算法

针对粒子滤波算法中粒子数自适应的问题,提出了一种新的算法.将当前滤波时刻的粒子随机划分为粒子数相同的两个粒子群,并采用对称KL距离方法计算他们之间的信息距离,然后根据信息距离的大小决定增加或者减少下一时刻参与滤波的粒子数,从而实现了滤波过程中粒子数目的自适应.该方法在确保一定滤波精度的基础上,能够减少滤波过程中需要的粒子数,为降低粒子滤波算法的时间复杂度提供了新的途径.仿真结果表明了算法的有效性.

An adaptive particle filter algorithm based on symmetrical KL distance

Aiming at the self adaption of the particle number in the particle filtering,a new algorithm is proposed.Firstly,the particles in current filtering time step are partitioned randomly to two particle swarms,and information distance between them is calculated by using symmetrical KL distance method.Subsequently,the increase or decrease of the particle number used in the next time step will be decided according to the information distance,and the self adaption of the particle number is obtained in the filtering.The proposed algorithm can reduce the particle number on the basis of the same filtering accuracy,and then it is a new approach that lowers the computational complexity in the particle filtering.Simulation results show the proposed algorithm is effective.