基于奇异值分解滤波的Prony技术在异步电机偏心故障检测中的研究  

Research on the Detecting Eccentricity Fault of Asynchronous Motor Based on Singular Value Decomposition Filter and Prony Technique

作　　者：张菁菁[1] 王臻[2] 

机构地区：[1]咸宁职业技术学院,湖北咸宁437000 [2]华中科技大学电气与电子工程学院,湖北武汉430074

出　　处：《电机与控制应用》2016年第3期83-88,共6页Electric machines & control application

摘　　要：针对异步电机偏心故障,提出一种新的利用奇异值分解滤波和Prony结合的检测技术。首先为了抑制背景噪声,对建立的Hankel矩阵进行奇异值分解(SVD),通过对较小奇异值置零滤除信号中的噪声。其次,为了提高检测精度,对滤除噪声后的信号进行Prony检测,准确估算出故障频率、幅值、谐波个数等参数。仿真和试验算例共同表明,SVD-Prony算法克服了传统Prony抗噪性差的缺点,具有快速傅里叶变换无可比拟的高分辨率和强抗噪能力,可在短时数据下达到检测偏心故障的目的。For eccentricity fault of asynchronous motor, a new detected method based on singular value decomposition filter and Prony Technique was proposed. First,in order to restrain background noise,the established Hankel matrix was processed by singular value decomposition,then noise was filtered by adjusting some smaller singular value to zero. second,for high detection precision,the fault signal without nosie was detected by Prony,and the fault frequency,amplitude and number of harmonic frequency were obtained. Simulation and experimental verification jointly showed that SVD-Prony overcomed the poor anti-noise of traditional Prony,whose advantages was quite outstanding than FFT in frequency resolution and restrain nosie. Simultaneous,SVD-Prony accurately detected eccentricity fault at short time data.

关 键 词：偏心故障 奇异值分解 PRONY 噪声 分辨率 

分 类 号：TM307[电气工程—电机]