基于小波包变换的转子断条故障特征提取

Extraction of Fault Characteristics of Rotor-Bar Breakage Based on Wavelet Packet Transform

针对鼠笼式感应电动机断条故障会在定子电流中产生特征频率的特点,采用双Hilbert变换对采集到的电流信号预处理,并且利用小波包变换提取故障特征信号。通过小波包分解,使得故障频率在每个子频段中突显出来。通过增加小波时域波形的波峰数,有效地抑制频带重叠现象和频谱泄露。双Hilbert变换解决了基频能量串扰的问题,让故障频率更容易提取。采用子频段节点重构系数均方根值变化率作为故障判断考察指标。通过实验室应用,验证该方法能够有效地识别转子断条故障。

In view of the appearing certainty of the characteristic frequency in the stator current during rotor-bar breakage fault of the induction motor, the collected current signal is pre-processed by DHT （Double Hilbert Transform） and then extracted the fault characteristic signal by WPT （Wavelet Packet Transform）. The WPT makes the appearance of fault frequency in individual sub-bands. The spectral leakage and the band overlap have been effectively suppressed by increasing the peak number of the waveform in the wavelet time domain. The energy crosstalk of the basic frequency is eliminated through DHT, which eases the extraction of fault frequency. The RMS （Root-Mean-Square） change rate of reconstruction coefficients of sub-band nodes are taken as the index for fault identification. The application to lab test verifies that the method is able to identify effectively the rotor-bar breakage fault.