摘要
针对粒子滤波算法存在计算量大和重采样过程带来计算量的扩张随粒子数的增加而成级数增加的问题,改进了一种基于重要性权值选择的粒子滤波方法.该方法运用重要性权值选择准则去除无效粒子而保留对状态估计贡献较大的粒子,每一状态时刻都根据权值选择的结果决定所要利用的粒子及其数目.仿真结果表明,该方法不仅保持了较高的滤波性能,而且能有效减小算法的计算量,有利于工程实现.
Aimed at the problem that the bulky computation and the resampling process making this calculation expand progressively with the increase of the particle numbers exist in the particle filter algorithm, an algorithm based on selection of importance weight was improved. By using the rule of selection of importance weight, those noneffective particles could be eliminated and the particles that contribute more to the state estimation were left. At each state the wanted particles and their number were determined in terms of the results of the weight selection. Simulation results show that the particle filter based on selection of importance weight can not only keep the better filtering performance but also effectively reduce filtering computation, which is propitious to the application in engineering.
出处
《空军雷达学院学报》
2009年第1期1-3,6,共4页
Journal of Air Force Radar Academy
关键词
重要性权值
粒子滤波
重要性采样
重采样
状态估计
importance weight
particle filtering
importance sampling
resampling
state estimation
