上海交通大学学报

上海交通大学学报（自然版） ›› 2011, Vol. 45 ›› Issue (10): 1526-1530.

• 自动化技术、计算机技术 • 上一篇    下一篇

基于遗传算法的改进粒子滤波算法

杨宁1，钱峰2a，朱瑞2b   

  1. (1. 上海电力学院 电力与自动化工程学院， 上海 200090； 2. 上海交通大学 a. 电子信息与电气工程学院， 上海 200240； b. 机械与动力工程学院， 上海 200240)
  • 收稿日期:2010-11-30 出版日期:2011-10-31 发布日期:2011-10-31
  • 基金资助:

    国家自然科学基金项目( 60801056), 上海市青年科技启明星计划(11QA1402800), 上海教委科研创新重点项目(11ZZ170)

Improved Particle Filter Based on Genetic Algorithm

 YANG  Ning-1, QIAN  Feng-2a, ZHU  Rui-2b   

  1. (1. School of Electric Power and Automatic Engineering, Shanghai University of Electric Power, Shanghai 200090, China; 2a. School of Electronic, Information and Electrical Engineering;2b. School of Mechanical Engineering, Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2010-11-30 Online:2011-10-31 Published:2011-10-31

摘要： 针对粒子数量和质量对粒子滤波(Particle Filter，PF)的退化问题具有重要影响，从大量采样粒子中采用遗传算法(Genetic Algorithm，GA)获得采样重要性重采样粒子滤波（Sampling Importance Resampling Particle Filter，SIRPF）的初始粒子，改善初始粒子质量，并保证其随机性和统计性.在车辆定位仿真中，采用定位精度、滤波发散次数和计算时间为指标对改进的遗传粒子滤波算法GASIRPF和传统SIRPF进行比较.结果表明，GA改进了初始粒子质量，减少了粒子退化可能性，提高了系统定位精度.

关键词:  粒子滤波, 遗传算法, 粒子质量, 定位

Abstract:  Considering particle number and particle quality have major influence on degeneracy phenomenon, a set of particles was chosen from a lot of initial sample particles using genetic algorithm, and kept the similarity and randominess, means and covariance as the initial sample particles. This method is used in a land vehicle navigation system, and compared with the tradition particle filter with the same amount of initial particles. The location precision, the degeneracy times and calculation time are used as the calibrating performances of the algorithms. The results prove that the initial particles quality is improved, the particle degeneracy phenomenon reduced and the system location precision improved.

Key words: particle filter, genetic algorithm, particle quality, location

中图分类号: