基于正交实验设计的粒子滤波算法  

Particle Filter Algorithm Based on Orthogonal Experimental Design

作　　者：文志强[1] 曾志高[1] 邓晓军[1] 

机构地区：[1]湖南工业大学计算机与通信学院,湖南株洲412007

出　　处：《小型微型计算机系统》2016年第7期1536-1541,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61170102)资助;湖南省自然科学基金项目(14JJ2115;2015JJ2046)资助

摘　　要：为了解决粒子退化问题,提出一种基于正交实验设计的粒子滤波算法.在粒子滤波算法中引入正交实验设计思想,利用构建的正交实验设计方案将粒子进行重新组合、粒子评估和最优选择,以改善粒子性能.该方法类似于遗传算法中的交叉和变异,但产生的新粒子更具有代表性.在数值仿真实验中,利用两个常用仿真模型,比较了提出算法与经典粒子滤波,进化粒子滤波算法的性能,探讨了相关参数对状态估计性能的影响.大量的实验结果表明提出的粒子滤波算法性能要好于其他四种粒子滤波算法.A particle filter algorithm based on orthogonal experiment design was presented to solve the problem of particle degeneracy.The idea of orthogonal experiment design was integrated into particle filter. The idea of orthogonal experiment design was integrated into particle filter algorithm. By using the constructed scheme of the orthogonal experiment design,the particles were recombined,assessed and optimally chosen to improve the performance. This method is similar to the cross operator and mutation operator in genetic algorithm,but the newparticles generated in our method are more representative than those in genetic algorithm. In numerical simulation experiment,two classic models were used to test the performance of our method against classic particle filter and evolution particle filter,and the effection of related parameters to the state estimation performance was discussed. A large number of comparison experiments showed that the performance of our particle filter algorithm is better than that of other four particle filter algorithms.

关 键 词：粒子滤波 正交实验设计 粒子退化 重采样 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]