基于SPWVD时频图纹理特征的滚动轴承故障诊断

针对如何提高滚动轴承故障诊断准确率的问题,提出一种基于平滑伪维格纳-威利分布(smooth and pseudo Wigner-Ville distribution,简称SPWVD)时频图纹理特征的故障诊断方法,对滚动轴承不同故障类型及故障程度进行识别。首先,采用SPWVD时频分析方法处理轴承故障振动信号,并获取时频图,从中提取选择表征能力优秀的特征参量作为故障特征;其次,将故障特征作为输入,结合支持向量机(support vectors machine,简称SVM)建立滚动轴承故障诊断模型;最后,采用轴承故障数据,比较SPWVD时频图纹理特征、维格纳-威利分布(Wigner-Ville distribution,简称WVD)时频图纹理特征和小波尺度谱图纹理特征3种故障特征的模式识别能力及准确率。分析结果表明,SPWVD时频图纹理故障特征分类效果最佳,敏感性最强,具有较高的故障诊断精度。

Rolling Bearing Faults Diagnostics Based on SPWVD Time-Frequency Distribution Image Texture Feature

In order to improve the accuracy of fault diagnosis for rolling bearings, this paper proposes an intelligent fault diagnosis method based on smooth and pseudo Wigner-Ville distribution (SPWVD) time-frequency distribution image texture features. First, we used the SPWVD time-frequency analysis method to process the bearing fault vibration signal. We acquired the time-frequency distribution image, then extracted the texture features from the images for the formation of the rolling bearing fault feature vectors based on certain sensitive texture features. Then, we introduced the fault feature vectors as the input to achieve fault diagnosis of rolling bearings based on the support vector machine (SVM). Finally, we extracted three kinds of feature vectors from the bearing fault data using the SPWVD time-frequency analysis method, Wigner-Ville distribution time-frequency analysis method, and wavelet scale spectrum method, respectively. We acquired the fault diagnosis accuracy of these feature vectors experimentally. The comparison results showed that the performance of the SPWVD time frequency texture fault feature outperformed the two other kinds of feature vectors with the best classification accuracy and sensitivity.