基于高频电流信号的电机故障特征提取方法

Motor Fault Feature Extraction Method Based on High Frequency Current Signal Demodulation

目前异步电机故障诊断主要依赖于振动、温度和噪声等参数,对于监测的电气信号,主要用于分析电机输出动力情况。电气信号中除了体现电机动力输出特征外,还包含丰富的机械故障、电气故障等信息特征。由于电气信号中的故障特征信号微弱,易被基频分量与噪声湮没而难以突出故障特征,不利于电机状态监测与故障诊断,因此提出了一种基于高频电流信号解调的电机故障特征提取方法,针对采样率不低于25.6 kHz的电机定子高频电流信号,综合运用高通滤波、Hilbert变换、快速傅里叶变换(FFT)3种分析方法,提取高频电流中电机故障的微弱特征信号,应用神经网络算法对电机正常、静态偏心、动态偏心、转子断条、轴承内圈、轴承外圈6种状态进行了诊断。该方法准确提取了电流信号故障特征,并成功识别了6种故障。

The asynchronous motor fault diagnosis mainly depends on the parameters such as vibration,temperature and noise,the monitoring of electrical signals,it is mainly used for analyzing the motor output power.In addition to the power output of the motor,the electrical signal also contains a wealth of mechanical fault,electrical fault and other information characteristics.Because the electrical signal in the fault characteristic signal is weak,easily lost to the fundamental frequency component and noise is to highlight the fault characteristic,is not conducive to machine condition monitoring and fault diagnosis.A method based on high frequency current signal demodulation of motor fault feature extraction method is proposed,in view of the sampling rate is not lower than 25.6 kHz high frequency current signal of motor stator,three analysis methods of high-pass filter,Hilbert transform and fast Fourier transform(FFT)are used to extract the weak characteristic signals of motor faults in high frequency current.The neural network algorithm is used to diagnose the motor normal,static eccentricity,dynamic eccentricity,rotor broken strip,bearing inner ring and bearing outer ring,and the effectiveness of the method is verified.The fault characteristics of current signal are extracted accurately and six kinds of fault are identified successfully.