半监督拉普拉斯分值在滚动轴承故障诊断中的应用  

APPLICATION OF SEMI SUPERVISED LAPLACE SCORE IN ROLLING BEARING FAULT DIAGNOSIS

作　　者：梁闯 陈长征[1] 刘野 贾歆莹 LIANG Chuang;CHEN ChangZheng;LIU Ye;JIA XinYing(School of Mechanical Engineering,Shenyang University of Technology,Shenyang 110870,China;BMW Brilliance Automotive Co.,Ltd.,Shenyang 110143,China)

机构地区：[1]沈阳工业大学机械工程学院,沈阳110870 [2]华晨宝马汽车有限公司,沈阳110143

出　　处：《机械强度》2023年第4期771-777,共7页Journal of Mechanical Strength

基　　金：国家自然科学基金项目(51675350)资助。

摘　　要：针对滚动轴承故障诊断过程中标签样本不足的问题,结合特征选择与二次挖掘,提出了基于半监督拉普拉斯分值(Semi Supervised Laplace Score,SSLS)和核主元分析(Kernel Principal Component Analysis,KPCA)的滚动轴承故障诊断模型。SSLS将半监督思想应用于拉普拉斯分值特征选择方法中,利用少量的有标签样本和大量无标签样本,结合KPCA对故障特征进行二次挖掘。同时,将粒子群优化的支持向量机(Particle Swarm Optimization-based Support Vector Machine,PSO-SVM)算法用于故障分类。最后,将该模型应用于实验数据分析过程。结果表明,该模型在减少样本标记工作量的同时,仍能在滚动轴承故障分类中保持较高的准确率,验证了所建立模型的有效性和工程实用性。Aiming at the problem of insufficient labeled samples in the process of rolling bearing fault diagnosis,a rolling bearing fault diagnosis model based on semi supervised Laplace score(SSLS)and kernel principal component analysis(KPCA)is proposed by combining with the idea of feature selection and secondary mining.SSLS applies the semi supervised idea to the Laplace score feature selection method,uses a small number of labeled samples and a large number of unlabeled samples,and combines KPCA to excavate fault features for a second time.At the same time,particle swarm optimization-based support vector machine(PSO-SVM)algorithm is used for fault classification.Finally,the model is applied to the process of experimental data analysis.The results show that the model can not only reduce the workload of sample marking,but also maintain a high accuracy in rolling bearing fault classification,which verifies the effectiveness and engineering practicability of the model.

关 键 词：特征选择 半监督拉普拉斯分值 核主元分析 粒子群优化的支持向量机 故障诊断 

分 类 号：TH165[机械工程—机械制造及自动化] TN911[电子电信—通信与信息系统]