基于相关向量回归和轨迹相似度分析的滚珠丝杠轴承剩余使用寿命预测  

Remaining Useful Life Prediction of Ball Screw Bearings Based on Relevance Vector Regression and Trajectory Similarity Analysis

作　　者：薛斌 许忠斌[2,3] 王鹏飞[3] 郑志功 XUE Bin;XU Zhongbin;WANG Pengfei;ZHENG Zhigong(School of Intelligent Manufacturing,Taizhou University,Taizhou 318000,China;College of Energy Engineering,Zhejiang University,Hangzhou 310027,China;School of Engineering,Hangzhou City University,Hangzhou 310015,China;Luoyang Bearing Research Institute Co.,Ltd,Luoyang 471039,China)

机构地区：[1]台州学院智能制造学院,浙江台州318000 [2]浙江大学能源工程学院,杭州310027 [3]浙大城市学院工程学院,杭州310015 [4]洛阳轴承研究所有限公司,河南洛阳471039

出　　处：《轴承》2025年第5期60-70,共11页Bearing

基　　金：国家自然科学基金面上项目(52073247)。

摘　　要：针对滚珠丝杠轴承的剩余使用寿命预测问题,提出了基于相关向量回归和轨迹相似度分析的剩余使用寿命预测方法。基于相关向量机训练得到在役轴承和退役轴承退化特征的相关向量,借助轨迹相似度计算给出在役轴承的预测参考段和参考相关向量,基于经验退化模型拟合相关向量得到轴承退化过程的代理模型,最后得到轴承退化趋势和剩余使用寿命预测结果。在加速寿命试验台上开展了滚珠丝杠轴承加速寿命试验,得到了4个轴承全生命周期振动信号数据。与同类算法的对比结果表明,该方法的平均预测误差最小,均方根和平均绝对值预测误差分别为6.12和5.57,预测精度高于其他算法。To predict the remaining useful life of ball screw bearings,a prediction method is proposed based on relevance vector regression and trajectory similarity analysis.The relevance vectors of degradation features of inservice and failed bearings are obtained based on relevance vector machine training.The prediction reference segments and reference relevance vectors of in-service bearings are given by trajectory similarity calculation.The surrogate model of bearing degradation process is obtained by fitting the relevance vectors using an empirical degradation model,ultimately yielding the degradation trend and remaining useful life prediction results of the bearings.The acceleration life test of the bearings is carried out on accelerated life test bed,and the vibration signal data for full life cycle of four bearings are obtained.The comparison with similar algorithms demonstrate that the proposed method achieves the smallest average prediction error,with root mean square and mean absolute prediction errors of 6.12 and 5.57,respectively,indicating higher prediction accuracy than other algorithms.

关 键 词：滚动轴承 滚珠丝杠 剩余寿命 预测 相关向量机 相似度 拟合模型 

分 类 号：TH133.3[机械工程—机械制造及自动化] TP277[自动化与计算机技术—检测技术与自动化装置]