组合导航系统的自适应MCMC信息融合算法  

Information Fusion Using Adaptive MCMC Method in Integrated Navigation System

作　　者：周健[1] 裴福俊[1] 崔平远[1] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100124

出　　处：《系统仿真学报》2009年第16期5120-5123,共4页Journal of System Simulation

基　　金：国家863计划项目(2006AA12Z305)

摘　　要：针对组合导航系统为高维非线性非高斯的特点,提出了一种在线实时调整粒子个数的自适应MCMC(Markov Chain Monte Carlo)粒子滤波算法。该算法利用基于KLD(Kullback-Leibler Distance)采样方法的自适应粒子个数调整算法在线调整MCMC粒子滤波过程中的粒子个数,利用预测粒子在状态空间中的分布情况来在线实时的确定下一次滤波迭代所需的粒子个数,从而有效减小算法的运算量,提高MCMC粒子算法的实时处理能力。最后,将该算法应用到了组合导航系统中进行了仿真研究。通过仿真结果可以看出,该算法在保持了MCMC粒子滤波算法的估计能力的同时,有效降低了算法的计算量,更适合于实际应用。The integrated navigation system is a nonlinear and non-Gaussian system. Based on these features, an Adaptive MCMC （Markov Chain Monte Carlo） Particle Filter was proposed, in which the real time amount of particles was adaptively adjusted online. The MCMC process in the Adaptive MCMC Particle Filter was used to maintain diversity of particles, avoiding sample degeneracy and impoverishment; and the KLD sampling approach was imported in order to adaptively choose the number of particles according to the state uncertainty, improving efficiency. In the last part, an experiment in SINS/GPS Integrated Navigation System was given. The experiment shows a great reduction of calculate amount and time when using Adaptive MCMC Particle Filter instead of standard MCMC Particle Filter with the same accuracy.

关 键 词：组合导航 MCMC粒子滤波 KLD采样 非线性非高斯 

分 类 号：V448.224[航空宇航科学与技术—飞行器设计]