基于权值优化的粒子滤波算法研究  

Particle Filter Algorithm Based on Weight Optimization

作　　者：杜正聪[1] 辛强[1,2] 邓寻 

机构地区：[1]攀枝花学院,四川攀枝花617000 [2]西华大学电气信息学院,成都610039

出　　处：《重庆师范大学学报（自然科学版）》2015年第3期124-129,共6页Journal of Chongqing Normal University:Natural Science

基　　金：四川省应用基础研究项目(No.2011JY0115)

摘　　要：为解决粒子滤波算法中存在的权值退化和实时性差的问题,提出了一种改进的权值优化组合粒子滤波算法(impWOPF),该算法通过对粒子权值设定门限Thershold,剔除权重小于Thershold的粒子,减少不必要的粒子运算,然后对小于粒子群权值均值的粒子进行权值优化组合,以增大小权值粒子的权值,保持了粒子多样性,提高了算法的实时性。仿真结果表明,该算法能够在保证估计精度的同时,有效降低重采样过程中的计算量,有利于实时信号的处理。To solve weights degradation and real-time problem in particle filtering algorithm, this paper presented an improved weight optimization particle filtering algorithm （imp-WOPF）, the algorithm settled the threshold for particle weight value of the newly generated, and eliminating particles of weights less than threshold. Then, using weighted linear combination between particle weights less than mean of the particle swarm weights and mean of the particle swarm weights before resampling, inhibiting sample depletion and improving the estimation accuracy. With ensuring the accuracy of the estimated, it reduces effectively the amount of computation in the process of the resampling. The simulation results showed the effectiveness of the algorithm.

关 键 词：粒子滤波 权值优化 门限 均方根误差 重采样 

分 类 号：TN957.51[电子电信—信号与信息处理]