多重信号分类法与扩展Prony结合的异步电机转子故障检测  

Detection for Broken Rotor Fault in Induction Motors Combining Multiple Signal Classification Algorithm with Extended Prony

作　　者：许伯强[1] 田士华 

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《华北电力大学学报（自然科学版）》2015年第6期1-7,23,共8页Journal of North China Electric Power University：Natural Science Edition

基　　金：国家自然科学基金资助项目(51277077)

摘　　要：提出了一种基于奇异值分解滤波技术的多重信号分类算法与扩展Prony算法相结合的异步电动机转子断条故障检测的新方法。奇异值分解滤波技术可以高效率地滤除电机定子电流信号中的基频分量与有色噪声,从而凸显转子断条故障特征频率分量;多重信号分类方法可在短时采样数据条件下准确计算故障特征分量的频率;扩展Prony方法则可以精确计算出各特征分量的幅值,弥补了多重信号分类算法无法求解幅值的不足。因此,将三者结合即可在短时采样信号条件下以高频率分辨力准确提取转子断条故障特征频率分量。对一台异步电机进行试验,结果表明:新方法简单、实用,效果理想。Based on singular value decomposition filtering,multiple signal classification and extended Prony algorithm,a new method for detecting broken rotor bar fault（ BRB） in induction motors is proposed. Firstly,Simulations of BRB show that the singular value decomposition filtering filter can efficiently filter the fundamental component and the color noises from motor stator current signals,which highlights the BRB fault feature component. Secondly,The algorithm of multiple signal classification can clearly identify the frequencies of broken rotors bar feature components even with short-time sampling data. Finally,extended Prony can accurately calculate the amplitudes of the feature components,which covers the shortage of multiple signal classification. By combining the three methods listed above,the study can extract the fault feature frequencies with high frequency resolution from short signals. Thus,the experimental results of an induction motor show that the method,based on singular value decomposition filtering,combing multiple signal classification with extended Prony,has a desirable performance.

关 键 词：异步电动机 转子断条故障 奇异值分解 多重信号分类 扩展Prony 

分 类 号：TM307[电气工程—电机]