矿用电动机振动信号早期故障特征提取方法

Early fault feature extraction method of vibration signal of mine-used motor

针对现有矿用电动机振动信号故障特征提取方法存在依赖参数设置、频率混叠、信号失真等问题,提出了一种基于双树复小波变换的矿用电动机振动信号早期故障特征提取方法。利用双树复小波变换对采集的矿用电动机振动信号进行分解,得到各层双树复小波系数,并采用软阈值滤波对各层双树复小波系数进行滤波处理,滤波处理后的双树复小波系数经双树复小波变换重构获得去噪信号。应用结果表明,该方法能有效去除电动机振动信号中噪声,提取的早期故障特征能很好地反映电动机实际运行工况,为电动机早期故障诊断提供了有效依据。

In view of problems of parameter setting,frequency aliasing and signal distortion existing in current fault feature extraction methods of vibration signal of mine-used motor,an early fault feature extraction method of vibration signal of mine-used motor based on dual-tree complex wavelet transform was proposed.Firstly,collected vibration signal of mine-used motor is decomposed by using dual-tree complex wavelet transform,so as to obtain dual-tree complex wavelet coefficients of each layer.Then soft threshold filtering is used to filter the dual-tree complex wavelet coefficients of each layer.At last,denoising signal is obtained by reconstruction of the filtered dual-tree complex wavelet coefficients.The application results show that the method can effectively remove noise in the motor vibration signal,and extracted early fault feature can reflect actual operating condition of motor,which provides an effective basis for early fault diagnosis of motor.