一种新的粒子滤波算法在INS/GPS组合导航系统中的应用

Improved particle filter algorithm for INS/GPS integrated navigation system

中文关键词  粒子滤波  最大期望算法  惯导系统/全球定位  组合导航

英文关键词  particle filter  expectation-maximization algorithm  INS/GPS  integrated navigation

中文摘要

      为改善传统粒子滤波中的样本退化和样本枯竭问题, 提出一种新的粒子滤波算法. 在重要性采样中, 利用最新测量值, 结合差分滤波算法产生重要性函数; 在再采样中, 利用高斯混合模型近似状态的后验概率密度, 引入最大期望算法计算该高斯混合模型的参数, 并从该新分布中采样后验粒子集, 取代传统的再采样. 从而通过提高重要性函数对状态后验概率密度的逼近程度来缓解样本退化问题, 通过改进再采样实现过程来缓解样本枯竭问题. 把新算法应用到INS/GPS组合导航系统中, 仿真结果表明新算法的估计性能明显优于粒子滤波.

英文摘要

      To deal with the problem “sample degeneration”and “sample impoverishment”in traditional particle filter(PF), a new particle filter algorithm is presented. In the importance sampling, this new filter uses the difference filter and the latest observed measurements to generate the importance-density. In the resampling step, it makes use of the posterior density of the approximate states in the Gaussian mixture model and employs the expectation-maximization(EM) algorithm to calculate the parameters of the Gaussian mixture model; and then, it samples the posterior particle sets from the new distribution to replace the traditional resampling step. The effects of the “sample degeneration”and the “sample impoverishment”are reduced. Simulation results of the application to inertial navigation system/global position system(INS/GPS) integrated navigation system show that the estimation performance of proposed algorithm is superior to that of the traditional particle filter algorithm.