摘要
目标被动追踪利用持续的观测信息来估计目标的运动状态,针对此问题提出了一种改进残差重采样粒子滤波算法.算法考虑采样粒子集的空间分布特性,将粒子集空间分布分割为数量可变、可数的网格,在每个网格内运用时间序列相关性分析选择重要粒子,能够丰富采样粒子的多样性,并将该网格内所有粒子的残余权值和赋予该重要粒子,从而削弱采样粒子的退化现象,提高非线性系统状态估计精度.实验表明:当观察噪声方差小于系统噪声方差,特别是当初始采样粒子数目较小时,该算法在单站纯方位目标追踪状态估计中的精度优于传统残差重采样粒子滤波算法.
Problems in single station passive target tracking for estimating the state of the moving target based on successive measurement have been studied. Based on this, an improved residual resampling particle filter algorithm was proposed, which prevents uncensored discarding of the low weighted particles and maintains the diversity of the sample particles. The key idea in the new algorithm is to select the important particles based on not only their weight but also their state values. Simulations of single station passive target tracking demonstrate the estimation accuracy of the algorithm, which is better than the traditional residual resampling method, especially when the sample size is small.
出处
《天津科技大学学报》
CAS
北大核心
2016年第6期74-78,共5页
Journal of Tianjin University of Science & Technology
基金
国家自然科学基金资助项目(51674176
81472070)
天津市高等学校科技发展基金资助项目(20130707)
关键词
粒子滤波
纯方位目标追踪
残差重采样
particle filter
bearing-only target tracking
residual resampling
