一种摄动粒子滤波故障检测方法

粒子滤波算法在故障检测应用中面临的两个难题是退化问题和难以跟踪突变状态。针对上述问题,将随机摄动再采样方法和强跟踪滤波算法引入粒子滤波,提出了一种摄动粒子滤波故障检测方法,旨在于解决粒子滤波的退化问题并提高算法对突变状态的跟踪能力,从而提高故障检测方法对故障的检测准确度。通过强跟踪滤波更新粒子,来提高算法跟踪突变状态的能力;当出现退化现象时,采用随机摄动再采样方法,对粒子集中的最优粒子迭加一个随机摄动量,用摄动粒子替换粒子集中的退化粒子,解决退化问题。仿真结果显示该算法能及时、准确地检测系统故障。

A particle filter with perturbation for fault detection

Degeneracy phenomenon and the ability to track breaking states are two difficult problems for particle filter applied to fault detection .A strong tracking particle filter with perturbation for fault detection was proposed to resolve the above problems ,in which ,strong tracking filter and stochastic perturbation re-sampling were introduced to advance the veracity of fault detection .Strong tracking filter was introduced to improve the ability to track breaking states ,and stochastic perturbation re-sampling was introduced to up-date the particles when the degeneracy phenomenon emerged .A stochastic perturbation was added to the best particle to form some new particles to displace the degenerative particles to solve the degeneracy phe-nomenon .The proposed algorithm accurately predicts faults in good time which is demonstrated by the simulation results .