基于PCANet_SVM的滚动轴承故障诊断

Fault Dignosis of Rolling Bearing Based on PCANet_SVM

针对传统滚动轴承故障诊断方法大多依赖于人工特征提取且故障诊断过程复杂导致泛化性能较差等问题,提出了一种基于主成分分析网络与SVM分类器相结合的滚动轴承故障诊断方法——PCANet_SVM。PCANet算法采用非监督方式进行特征提取,与SVM分类器结合进行故障识别。实验结果表明,PCANet_SVM算法可以有效地实现滚动轴承故障诊断,通过与CNN算法进行对比,验证了本算法的高效性和准确性。

Aiming at the problems that traditional rolling bearing fault diagnosis methods mostly rely on manual feature extraction and the complex fault diagnosis process leads to poor generalization performance,a new rolling bearing fault diagnosis method based on the combination of principal component analysis network and SVM classifier is proposed-PCANet_SVM.The PCANet algorithm uses an unsupervised method for feature extraction and combines with the SVM classifier for fault identification.The experimental results show that the PCANet_SVM algorithm can effectively realize the fault diagnosis of rolling bearings.By comparing with the CNN algorithm,the efficiency and accuracy of this algorithm are verified.