基于改进粒子滤波的非线性系统故障诊断

Fault Diagnosis of Nonlinear Systems Based on Improved Particle Filter

粒子滤波在非线性系统的故障诊断中,存在着粒子样本退化和突变状态难以跟踪的问题。为此提出一种基于无迹变换和遗传变异改进的粒子滤波算法,通过无迹变换将粒子转移到高似然区域,遗传算法代替重采样消除粒子多样性退化的问题,再利用对数似然函数和作为评价指标来进行故障诊断。仿真实验结果表明,改进的算法可有效提高滤波精度,在连续搅拌反应器变量发生突变时,能够有效、准确地诊断出故障。

Particle filter is used in the fault diagnosis of nonlinear systems, there is the problem that the particle samples are degraded and it is difficult to track the mutation state. Therefore, a particle filter algorithm based on unscent transformation and genetic variation improvement is proposed. Through the unscent transformation, the particles are transferred to the high-likelihood region. The genetic algorithm replaces resampling to eliminate the degradation of particle diversity, and then the log-likelihood function is used as an evaluation index to perform the fault diagnosis. The simulation experiment results show that the improved algorithm can be used to effectively improve the filtering accuracy, and effectively, accurately diagnose the fault when the mutation occurs in the variable of the continuous stirred reactor.