作　　者： (周建民）; (郭慧娟）; (张龙）;

机构地区： 华东交通大学机电与车辆工程学院,江西南昌330013

出　　处： 《华东交通大学学报》 2017年第4期110-116,共7页

摘　　要： 滚动轴承的运行状态与退化程度息息相关,若能对轴承的退化程度进行在线定量评估,则可使设备维护策略的制定具有针对性。本文对无故障样本进行小波包分解得到能量值并将其和时域值作为原始特征。对原始特征进行降维后分为训练和待测数据,用无故障样本训练HMM模型,稳定后保持模型不变通过迭代的方式将待测样本输入到训练好的HMM,获得最大输出似然概率作为性能退化程度指标,用轴承加速疲劳试验和包络解调对本文的结论进行验证。本文提出的性能退化方法得到的结论与轴承加速疲劳试验得到的结果一致。 The running condition of rolling bearing is closely related to the degree of its degradation. If the degradation degree of rolling bearings can be assessed online quantitatively, the equipment maintenance strategy will be pertinent. This paper makes the node energy values decomposed of non-faulty samples by wavelet packet,which are taken together with time-domain features as the original characteristics of the signals. The original features are classified into training data and test data after the nonlinear flow-based dimensionality reduction.The HMM model is trained by the non-faulty samples. After the model is stabilized and the model is maintained unchanged, samples to be tested are input into the trained HMM through iterations. Then, the maximum output likelihood is obtained as performance degradation index, which is adopted to evaluate the performance of rolling bearings. The proposed method is verified by fatigue life test of the bearing and the envelope demodulation. Results of performance degradation method are in agreement with those obtained from the accelerated fatigue tests of bearings.

关 键 词： 隐马尔科夫 滚动轴承 小波包分解 时域特征 性能退化评估 包络解调