摘要
如何得到重要性密度函数是粒子滤波算法的关键问题之一。首先阐述了粒子滤波的一般方法;然后在分析修正无偏量测转换统计特征的基础上,提出了一种修正无偏量测转换粒子滤波(M-UCMPF)算法,推导了该算法的重要性概率密度函数;最后通过仿真实验对比分析了M-UCMPF算法、不敏卡尔曼粒子滤波(UPF)算法、修正无偏量测转换卡尔曼滤波(M-UCMKF)算法和不敏卡尔曼滤波(UKF)算法的性能。仿真结果表明,M-UCMPF算法具有运算量相对较小而滤波精度高的特点。
How to get the importance density function is one of the key problems of partical filter.The principle of particle filter is given at first.Based on the analysis to statistical characteristics of the Modified Unbiased Converted Measurement,a Modified Unbiased Converted Measurement Particle Filter（M-UCMPF） is proposed,and its importance density function is derived.Through simulation experiments,a comparison is made for the performance of four algorithms： M-UCMPF,Unscented Particle Filter（UPF）,Modified Unbiased Converted Measurement Kalman Filter（M-UCMKF） and Unscented Kalman Filter（UKF）.The simulation results indicate that M-UCMPF has relatively small computational cost and higher filtering accuracy for the target tracking.
出处
《电光与控制》
北大核心
2010年第5期30-34,共5页
Electronics Optics & Control
基金
中电科技集团第14研究所所控课题
关键词
无偏量测转换
粒子滤波
重要性密度函数
unbiased converted measurements
particle filter
importance density function
