论文部分内容阅读
粒子滤波算法通过非参数化蒙特卡罗仿真方法实现递推贝叶斯滤波,基于序贯重要性采样的粒子滤波算法无法避免粒子退化问题;通过在滤波初始化阶段对初始化粒子进行优化选择,在重采样阶段使用非排序的基于权重的重采样算法对粒子滤波算法进行了改进,从一定程度上解决了粒子退化问题;仿真验证,本算法在保持与传统粒子滤波算法运算时间的条件下,提高了粒子滤波算法的估计精度,从而提高了其在机动目标跟踪中的性能。
Particle filter algorithm through non-parametric Monte-Carlo simulation method to achieve recursive Bayesian filtering, particle filter based on sequential importance sampling algorithm can not avoid the problem of particle degeneration; initial stage of the filter initialization optimization of particle selection, in the heavy In the sampling phase, the particle filtering algorithm is improved by using non-ordered weight-based resampling algorithm, to a certain extent, to solve the problem of particle degeneration. The simulation results show that this algorithm improves the performance of the traditional particle filter algorithm The estimation accuracy of the particle filter algorithm improves its performance in maneuvering target tracking.