独立分量分析在感应电动机转子故障特征提取中的应用  

Feature Extracting Method for Rotor Fault of Induction Motor Based on Independent Component Analysis

作　　者：方芳[1] 杨士元[1] 侯新国[2] 

机构地区：[1]清华大学自动化系,北京100081 [2]海军工程大学电气与信息学院,武汉430033

出　　处：《数据采集与处理》2007年第4期496-500,共5页Journal of Data Acquisition and Processing

摘　　要：在定子电流信号的频谱分析诊断感应电动机的故障时,转子断条故障特征频率分量常常被电流的基频分量淹没。针对这一情况,本文提出将独立分量分析应用于提取异步电动机转子故障特征。电流信号的信息空间是由电流的自相关矩阵Ri的特征向量张成的。将最大主分量(主要由基频分量构成)和噪声对应的特征向量去掉,由其余的特征向量张成降维信号子空间S*。将Ri投影到S*,削去Ri中的基频分量和噪声后,再利用独立分量分析提取故障特征分量。仿真和实验表明,该方法用于提取转子断条故障特征是可行并且是有效的。The rotor fault feature c behind the strong supply frequency omponent is difficult to component in the spect based on independent component analysis is proposed to e be detected because it always hides rum of the current. A novel method xtract the fault feature component. The eigenvectors of Ri, which is the autocorrelation matrix of the current i, span the information space of the current. Cut off the eigenvectors related to the noise and the supply frequency component, then the rest eigenvectors span the signal subspace S^＊. Project Ri to S^＊ , then the supply frequency component and the noise in Ri are taken out. After the pretreatment, the feature of the rotor broken-bar fault is picked out from R, by using independent component analysis. Simulation results and experiment show that the proposed method is feasible and effective.

关 键 词：感应电动机 独立分量分析 故障诊断 转子断条故障 信号子空间 

分 类 号：TM343[电气工程—电机]