摘要
粒子滤波存在的一个主要问题就是粒子的退化现象,虽然重采样可以有效缓解退化问题,但同时带来了粒子枯竭问题。为了解决上述问题,将权值选优的思想应用于粒子群优化的粒子滤波(PSOPF)。每次选择权值较大的部分粒子进行滤波,并且加入改进的多样性引导机制来保证粒子的多样性,提出一种双重优化的粒子滤波(DOPF)。仿真表明,改进算法相比基于权值选择的粒子滤波(WSPF)以及粒子群优化粒子滤波都在精度上有所提高,能克服WSPF存在的不足,并能在一定程度上应对状态大幅度突变情况,并提高了数据的精度。
A major problem of the particle filter is sample degeneration phenomenon. Although resample method can relieve the sample degeneration, at the same time, it brings about sample impoverishment problem. To solve this problem, the way of weight selected was used for PSOPF. These particles of larger weight for filtering were selected, a modified diversity formula was used to guarantee the diversity of particle, and then a double optimized particle filter was put forward. Simulations show that the DOPF can improve the precision compared with WSPF and PSOPF, over- come the flaw of WSPF and reply to mutation station of large range.
出处
《计算机仿真》
CSCD
北大核心
2014年第7期247-250,共4页
Computer Simulation