摘要
针对静电探测的数学模型结构复杂、强非线性以及实验测量数据存在极大不确定性的特点和传统粒子滤波(PF)算法存在的缺陷,提出了一种改进的粒子滤波(UPF)算法。该算法以无迹卡尔曼滤波(UKF)算法生成替代分布并从中采样,理论分析与仿真结果均表明,UPF算法能够提高静电探测系统目标跟踪的稳定性和精确性,解决了传统PF算法中以转换先验密度函数作为替代分布所引发的各种问题,具有较高的实用价值和广泛的应用前景。
Against the complexity and nonlinear of Electrostatic target tracking, uncertainty of the experimental data and the shortcomings of the traditional particle filter （PF）, an improved particle filter algorithm called un- scented particle filter （UPF） is proposed. This algorithm generates a proposal distribution by unscented Kalman filter （UKF） method and draws samples from it, the theoretical analysis and simulation results demonstrate that the UPF algorithm improves stability and accuracy of tracking, and solves the various problems caused by PF which uses transition prior density function as the proposal distribution. The UPF has high practical value and broad application prospect.
出处
《探测与控制学报》
CSCD
北大核心
2010年第2期64-67,72,共5页
Journal of Detection & Control
基金
山西省青年自然科学基金项目资助(2009021022-2)
关键词
静电探测
无迹粒子滤波
目标跟踪
electrostatic detection
unscented particle filter
tracking
