双树复小波与宽度学习在轴承故障诊断的应用

Fault Diagnosis of Bearing Based on Double Tree Complex Wavelet and Broad Learning System

针对滚动轴承时域信号难以有效提取其故障特征,且信号频谱在高低频区域内较为存在对分类无意义的冗余特征使得故障分类模型在训练过程中做无用功的问题,提出使用双树复小波进行故障特征提取。在此基础上,将双树复小波和宽度学习模型结合,提出了基于双树复小波与宽度学习的滚动轴承故障诊断方法。首先,利用双数复小波将采集到的振动信号分解为不同频带的子信号;然后提取子频带作为特征向量;最后用宽度学习对样本进行训练以完成快速故障分类。

For the issue that it is difficult to effectively extract fault characteristics of rolling bearings based on their time-domain signals and the redundancy characteristics of the signal spectrum insignificant to classification in areas with high and low frequencies make the fault classification model reinvent the wheel during training,it was proposed to employ double tree complex wavelet to extract fault characteristics.On this basis,double tree complex wavelet and broad learning system models were used together to propose a fault diagnosis method of rolling bearings based on double tree complex wavelet and the broad learning system.First of all,the vibration signals collected by double tree complex wavelet are divided into sub-signals at different frequency bands.Then sub-frequency bands are extracted as eigenvector.At last,samples are trained by the broad learning system to quickly complete fault classification.