基于多维退化特征与GG模糊聚类的滚动轴承退化状态识别  

Recognition of Rolling Bearing Degradation Condition Based on Multidimensional Degradation Feature and GG Fuzzy Clustering

作　　者：王微[1] 胡雄[1] 王冰 孙德建[1] WANG Wei;HU Xiong;WANG Bing;SUN Dejian(Logistics Engineering College, Shanghai Maritime University, Shanghai 201306, China)

机构地区：[1]上海海事大学物流工程学院

出　　处：《东华大学学报（自然科学版）》2019年第4期576-582,共7页Journal of Donghua University(Natural Science)

基　　金：国家高技术研究发展计划(“863”计划)资助项目(2013AA041106)

摘　　要：考虑滚动轴承性能退化状态在时间尺度上的连续性,将时间参数映射到指数函数中,形成更符合性能退化过程的弯曲时间(curved time,CT)参数,同时将C0复杂度和有效值(root mean square,RMS)分别作为复杂性维度和能量维度的退化特征,构建描述滚动轴承性能退化过程的三维特征向量[C0,RMS,CT]。在此基础上,采用GG(Gath-Geva)模糊聚类方法对滚动轴承性能退化状态进行阶段划分,识别不同的退化状态,选用分类系数、平均模糊熵以及序列离散度对聚类效果进行综合评价。采用来自IMS(intelligent maintenance system)的轴承全寿命试验数据进行实例分析,结果表明,提出的三维特征向量既能够反映滚动轴承性能退化趋势,又能体现同一状态在时间尺度上的连续性。Taking the continuity of rolling bearing degradation condition on time scale into consideration,we mapped time parameter into exponential function to form the curved time(CT)which was more conform to the performance degradation process.On this basis,a three-dimensional eigenvector was constructed including C0 complexity,RMS(root mean square)and three-dimensional vector[C0,RMS,CT].The GG(Gath-Geva)fuzzy clustering method was used to classify the performance degradation condition and identify different degradation conditions.The classification coefficient,average fuzzy entropy and the proposed sequence dispersion were used to evaluate the clustering effect comprehensively.The bearing life test data from IMS(intelligent systems division)was used for analysis.The results show that the proposed three-dimensional feature vector can reflect the performance degradation trend and the continuity for the same condition on the time scale.

关 键 词：C0复杂度 GG模糊聚类 滚动轴承 退化特征 

分 类 号：TH215[机械工程—机械制造及自动化] TH113.1