滚动轴承故障程度评估的AR-GMM方法  

Assessment of Rolling Bearing Fault Degree Using AR-GMM

作　　者：龙铭[1] 文章[1] 黄文艺[1] 周建民[1] 周继慧[1] 

机构地区：[1]华东交通大学机电与车辆工程学院,南昌330013

出　　处：《机械科学与技术》2016年第8期1183-1188,共6页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金资目(51265010;51205130);江西省自然科学基金项目(20161BAB216134);载运工具与装备教育部重点实验室项目(15JD02)资助

摘　　要：提出了一种基于AR-GMM的滚动轴承故障程度评估方法,该方法利用自回归模型(AR)提取无故障轴承早期振动信号特征,并建立无故障轴承高斯混合模型(GMM)作为故障程度评估基准。轴承后期振动信号在提取AR特征后导入该基准GMM模型,得到测试样本与无故障样本之间的量化相似程度。进而以此相似程度值为基础建立自回归对数似然概率值(ARLLP)作为滚动轴承故障程度评估指标。轴承疲劳试验分析表明该指标能够及时有效发现轴承早期故障,并能很好预测跟踪轴承恶化趋势,为视情维修奠定基础。A new method called AR-GMM is proposed based on autoregressive model（ AR） and Gaussian mixture model（ GMM） for fault degree assessment. Bearing vibration signals are represented by the parameters and residual of the AR model. A GMM is then obtained by training data from bearing fault-free. The data to be tested are fed to the baseline GMM to measure the similarity between the tested bearing condition and normal ones. Consequently,a bearing fault degree assessment indicator called ARLLP（ autoregressive log-likelihood probability） is formulated based on the similarity measure. Experiment results on bearing fatigue test demonstrate that the proposed method can detect bearing fault at early stage and can track the trend of deterioration of rolling bearings.

关 键 词：故障程度评估 视情维修 高斯混合模型(GMM) 

分 类 号：TH165.3[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]