基于构建退化特征和PSO-PF的滚动轴承寿命预测  

Rolling Bearing Life Prediction Based on Constructed Degradation Characteristics and PSO-PF

作　　者：安冬[1] 崔志强 邵萌[1] 李新然 王赛男 AN Dong;CUI Zhi-qiang;SHAO Meng;LI Xin-ran;WANG Sai-nan(School of Mechanical Engineering,Shenyang Jianzhu University,Shenyang 110168,China)

机构地区：[1]沈阳建筑大学机械工程学院,沈阳110168

出　　处：《组合机床与自动化加工技术》2023年第3期118-121,126,共5页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金(51975130,51975388)。

摘　　要：针对使用常规粒子滤波(PF)解决轴承寿命预测的方法中存在预测结果准确度不高的问题,提出了一种基于构建退化特征和粒子群优化粒子滤波算法(PSO-PF)的滚动轴承寿命预测方法。首先,将轴承振动信号经CEEMDAN-小波包联合去噪后提取多种时、频域特征;其次,经过综合指标评价与弗雷歇距离分析得到优秀退化特征集,用加权拟合的方法构建出新的退化特征;然后,通过特征的梯度变化来自适应检测轴承退化起始时间;最后,用PSO-PF算法对寿命预测模型进行更新来完成寿命预测。将构建的退化特征和PSO-PF算法预测结果分别与原始特征和PF算法预测结果进行对比,结果表明该方法更适用于轴承的寿命预测。In order to solve the problem of low accuracy of bearing life prediction using conventional particle filter(PF),a rolling bearing life prediction method based on constructed degradation features and particle swarm optimization particle filter algorithm(PSO-PF)was proposed.Firstly,the bearing vibration signals were denoised by CEEMDAN-wavelet packet,and a variety of time and frequency domain features were extracted;Then,excellent degradation feature sets were obtained by comprehensive index evaluation and Fréchet distance analysis,and new degradation features were constructed by weighted fitting method;Secondly,adaptive detection of bearing degradation start time through the gradient change of features;Finally,the life prediction model is updated by PSO-PF algorithm to complete the life prediction.The prediction results of degradation features and PSO-PF algorithm were compared with those of original features and PF algorithm,respectively.The results show that the proposed method is more suitable for bearing life prediction.

关 键 词：滚动轴承 寿命预测 退化特征 粒子群优化粒子滤波 

分 类 号：TH133.33[机械工程—机械制造及自动化] TG506[金属学及工艺—金属切削加工及机床]