基于融合模糊C均值与隐马尔科夫模型的滚动轴承的退化状态识别  

Degeneration State Recognition of Rolling Bearing Based on FCM-HMM Model

作　　者：周建民[1] 张臣臣 张龙[1] 郭慧娟[1] ZHOU Jianmin;ZHANG Chenchen;ZHANG Long;GUO HuiJuan(Key Laboratory of Conveyance and Equipment of Ministry of Education, East China Jiaotong University,Nanchang 330013,China)

机构地区：[1]华东交通大学载运工具与装备教育部重点实验室

出　　处：《机械设计与研究》2019年第3期83-86,共4页Machine Design And Research

基　　金：国家自然科学基金资助项目(51865010;51665013)

摘　　要：滚动轴承在长期的工作过程中其性能会出现不同程度的退化,如果能对滚动轴承的退化状态进行识别就可以做好维护措施。用自回归模型(Autoregressive model, AR)对滚动轴承全寿命周期的振动信号提取其系数及残差,用正常样本和失效样本特征建立模糊C均值模型(Fuzzy C Mean, FCM),用轴承正常样本的特征数据建立隐马尔科夫(Hidden Markov model, HMM)模型,将轴承的测试样本信号输入建立的FCM和HMM模型得到的两个退化指标,再将其作为特征矩阵输入到FCM模型,得到融合方法的性能退化曲线,结果表明该方法集中了空间统计距离模型和概率统计模型两者的优势,最后用IEEE PHM2012实验数据进行验证,表明所述方法与滚动轴承性能退化趋势保持一致并且可以提早发现早期故障。The performance of Rolling bearings will appear degradation of different degrees which run for a long time in the service period. If the degradation state of the rolling bearing can be identified we can do maintenance measures. The paper extracts the coefficients and residuals of the vibration signals of the full life cycle of rolling bearings using autoregressive model(AR). The Fuzzy C Mean model(FCM)is established using the normal and failure samples and the Hidden Markov model(HMM)is established using the normal samples. The two degradation indicators which was obtained by imputing the under test data to FCM and HMM model are input to the FCM model as the input characteristic. Then the performance degradation curve is obtained. The method combines the advantages of spatial statistical distance model and probabilistic statistical model. Then the IEEE PHM2012 experimental data are used to verify the conclusions of this paper. The experimental analysis shows that the method is consistent with the performance degradation trend of rolling bearings and can detect early faults early.

关 键 词：滚动轴承 AR模型 模糊C均值 隐马尔科夫模型 退化状态 IEEE PHM2012实验 

分 类 号：TH133[机械工程—机械制造及自动化]