非线性系统测量数据丢失时的一种粒子滤波器算法  

A Particle Filter Algorithm in the Presence of Missing Measurements for a Nonlinear System

作　　者：李雄杰[1,2] 周东华[2] 

机构地区：[1]浙江工商职业技术学院机电工程系,浙江宁波315012 [2]清华大学自动化系,北京100084

出　　处：《兵工学报》2009年第10期1405-1408,共4页Acta Armamentarii

基　　金：国家自然科学基金项目(60574084)

摘　　要：针对在工程实践中发生的测量数据随机丢失情况,提出了一种应用于非线性系统的滤波方法,该方法将基于序贯重要性采样的粒子滤波器应用于非线性、非高斯系统状态的在线状态估计。首先将测量数据丢失描述成满足一定条件概率分布的二元开关序列;然后基于似然函数设计方法,设计出测量数据丢失时的粒子滤波器算法;最后用本文方法对倒立摆系统状态估计进行了仿真。仿真实验表明,测量数据丢失时的粒子滤波器算法是有效的。Aimed at the case that sensor data may be missing randomly in practice, a filtering approach was proposed for the nonlinear systems, which applies a particle filter based on sequential importance sampling to the on-line state estimation of non-Gauss and nonlinear systems. The missing sensor data were described as a binary switching sequence which satisfies a certain conditional probability distribution; a particle filter algorithm in the presence of missing sensor data was designed based on likelihood function; the state estimation of a upside-down pendulum system was simulated by the proposed approach. The simulated results show the effectiveness of the proposed algorithm.

关 键 词：自动控制技术 粒子滤波器 测量数据丢失 非线性系统 状态估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]