滚动轴承故障程度评估的AR-GMM方法

发布时间：2018-05-19 16:49

  本文选题：故障程度评估 + 视情维修　； 参考：《机械科学与技术》2016年08期

【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。
[Abstract]:A fault degree evaluation method for rolling bearings based on AR-GMM is proposed. The autoregressive model (ARM) is used to extract the early vibration signals of fault free bearings, and a hybrid Gao Si model for fault free bearings is established as a benchmark for fault degree evaluation. After extracting the AR feature, the vibration signal of the bearing is imported into the benchmark GMM model, and the quantitative similarity between the test sample and the fault free sample is obtained. On the basis of the similarity value, an autoregressive logarithmic likelihood probability (ARLLP) is established as an index to evaluate the fault degree of rolling bearing. The analysis of bearing fatigue test shows that this index can detect the early failure of bearing in time and effectively, and can predict and track the deterioration trend of bearing well, and lay the foundation for maintenance according to the situation.
【作者单位】： 华东交通大学机电与车辆工程学院;
【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
【分类号】：TH133.33
，

本文编号：1910888