等距离映射和模糊C均值的滚动轴承故障识别  

Rolling Bearing with Isometric Feature Mapping and Fuzzy C-means Fault Identification Method

作　　者：王亚萍[1] 李士松 葛江华[1] 许迪[1] 李云飞 WANG Ya-ping;LI Shi-song;GE Jiang-hua;XU Di;LI Yun-fei(School of Mechanical and Power Engineering,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学机械动力工程学院

出　　处：《哈尔滨理工大学学报》2019年第3期41-47,共7页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金资助(51575143);黑龙江省自然科学基金资助(E2016046)

摘　　要：在滚动轴承的故障识别中,针对传统的等距离映射ISOMAP算法存在测地距离的计算偏差较大,故障识别部分混叠的问题,提出一种模糊C均值和等距离映射的滚动轴承故障识别方法。首先,对ISOMAP算法中的邻域大小k值用残差进行改进,保证映射结果很好地反映全局性质;其次可分性评价指标评价特征降维的效果;然后,采用了模糊C均值聚类方法,保证在拓扑空间中高维流形数据与低维空间光滑流形中的数据仍保持相近或相同的特性。最后,通过采集不同损伤程度下的滚动轴承振动数据进行实验验证,结果表明本文方法在分类效果和识别精度都有了明显的提升。In the fault identification of rolling bearing,the traditional ISOMAP algorithm is met with the problem of large deviation of geodesic distance and aliasing in fault identification.So,this paper presents a fuzzy C-means and Isometric Feature Mapping of rolling bearing fault identification method.First of all,the neighborhood size k of ISOMAP algorithm is improved with residuals to ensure that the mapping results reflect the global nature well.Second,the index of category divisibility is used to evaluate the effect of feature dimensionality reduction.Then,a fuzzy C-means clustering method is adopted to ensure that the data in high-dimensional manifolds and the low-dimensional smooth manifold in the topological space are still close or the same.Finally,the experimental verification of vibration data of rolling bearing with different damage degrees shows that the combination of fuzzy C-means and improved ISOMAP has obvious improvement in both classification and identification accuracy.

关 键 词：滚动轴承 故障识别 特征降维 ISOMAP算法 模糊C均值 

分 类 号：TH165[机械工程—机械制造及自动化] TN911.7[电子电信—通信与信息系统]