基于能量峰定位的经验小波变换及在轴承微弱故障诊断中的应用

Empirical Wavelet Transform Based on Energy Peak Location with Applications to Bearing Weak Fault Diagnosis

为了解决经验小波变换在轴承振动信号频谱分割不当的问题,提出了一种基于自适应能量峰定位的经验小波变换。该方法采用Teager能量算子对傅里叶频谱进行能量集中,降低噪声和无关分量影响;利用多尺度寻峰定位算法自适应确定频谱分割边界;通过构建小波滤波器组提取各模态分量。依据峭度指标挑选故障信息最大的模态分量,最后通过Hilbert包络解调提取轴承的故障特征频率。仿真和实验分析结果表明:提出的方法从能量角度入手,鲁棒性更强;频段划分考虑频谱的形状,能自适应识别故障频带;与原始经验小波变换方法相比,改进方法能明显增强早期微弱故障特征,提高轴承早期故障诊断性能。

To solve the problem of improper segmentation in vibration signal spectrum by empirical wavelet transform,an empirical wavelet transform based on the adaptive energy peak location is proposed,where Teager energy operator is used to concentrate the energy of Fourier spectrum to reduce the influence of noise and independent components.The spectrum segmentation boundary is determined adaptively by multi-scale peak-finding algorithm.Modal components are extracted by the constructed wavelet filter bank.According to kurtosis index,the modal component with maximum fault information is selected.The fault feature frequency of the bearing is extracted by Hilbert envelope demodulation.The simulation and experiment show that the proposed method is more robust from the perspective of energy.Considering the shape of frequency spectrum,fault frequency band can be identified adaptively.Compared with the original EWT method,this proposed method enables to obviously enhance the early weak fault features and improve the early fault diagnosis performance of bearings.