摘要
针对粒子滤波算法在故障预报中的大计算量和粒子退化问题,提出一种基于随机摄动粒子滤波器的故障预报算法.当粒子退化严重时,对粒子用随机摄动方式进行再采样,一方面可改进样本的多样性,缓解粒子退化;另一方面可缩短再采样时间,减少计算量,从而提高粒子滤波算法的跟踪能力.仿真结果表明该算法可行,能及时准确地对系统故障进行预报.
Degeneracy of particles and large computing cost are the main problems when particle filters are applied to fault predictions. Therefore, a fault prediction algorithm based on stochastic perturbation particle filter is proposed to resolve the above problems. The stochastic perturbation re-sampling is used when the degeneracy of particles is serious, which can improve the diversity of samples, ameliorate the degeneracy of particles, shorten the re-sampling time and reduce the computing cost. As a result, the.tracking ability of particle filter is improved. Simulation results demonstrate that the algorithm proposed is valid and the system fault can be predicted accurately and timely.
出处
《控制与决策》
EI
CSCD
北大核心
2009年第2期284-288,共5页
Control and Decision
基金
国家自然科学基金重点项目(60736026)
教育部新世纪优秀人才支持计划项目(NCET-07-0144)
关键词
粒子滤波
退化现象
计算量
随机摄动
故障预报
Particle filter
Degeneracy phenomenon
Computing cost
Stochastic perturbation
Fault prediction
作者简介
张琪（1980-），女，甘肃庆阳人，博士生，从事滤波理论、故障诊断和故障预报等研究；Correspondent： ZHANG Qi, E-mail zhangqi6530@ 163. com
胡昌华（1966-），男，湖北罗田人，教授，博士生导师，从事控制系统故障诊断、容错控制等研究．
