基于RMS子线区间拟合法的轴承剩余寿命预测  

Bearing Residual Life Prediction Based on RMS Sublinear Interval Fitting Method

作　　者：王贡献[1] 张腾 胡志辉[1] 杨仲 王绪光 李帅琦 WANG Gongxian;ZHANG Teng;HU Zhihui;YANG Zhong;WANG Xuguang;LI Shuaiqi(School of Transportation and Logistics Engineering,Wuhan University of Technology,Wuhan 430063,China;Wuhan K-crane Ocean Lifting Technology Co.,Ltd.,Wuhan 430063,China)

机构地区：[1]武汉理工大学交通与物流工程学院,武汉430063 [2]武汉开锐海洋起重技术有限公司,武汉430063

出　　处：《噪声与振动控制》2024年第6期165-171,共7页Noise and Vibration Control

基　　金：海南省自然科学基金资助项目(622MS097);国家重点研发计划资助项目(2022YFB2602303)。

摘　　要：轴承剩余使用寿命(Residual life,RL)预测模型的预测精度在很大程度上取决于衰退特征的趋势一致性。由于轴承个体差异性,导致同类型不同轴承的相同特征衰退趋势表现不同,从而导致训练集轴承建立的RL预测模型与测试集轴承不匹配。针对上述问题,本文提出一种基于振动信号均方根(RMS)趋势一致性的子线区间拟合法轴承寿命预测方法。依据μ±3σ原则提出一种奇异值替换方法,有效增强RMS衰退曲线的趋势一致性,并基于RMS衰退曲线高度的趋势一致性提出子线区间拟合法,利用XJTU-SY数据集对提出方法进行测试与验证,并与常见的寿命预测方法反向传播神经网络(Back Propagation Neural Network,BPNN)与长短期记忆神经网络(Long Short-term Memory Network,LSTM)预测结果进行对比,可有效提升轴承RL预测精度。The prediction accuracy of the residual life(RL)prediction model of bearings depends largely on the trend consistency of the recession characteristics.Due to the individual variability of bearings,different bearings of the same type exhibit different recession trends of the same characteristics,which leads to the mismatch between the RL prediction model established for the training set bearings and the test set bearings.Aiming at the above problems,this paper proposed a bearing life prediction method based on the trend consistency of the root-mean-square(RMS)of vibration signals for the subline interval fitting method.According to the principle of μ±3σ,a singular value replacement method was proposed to effectively enhance the trend consistency of the RMS recession curve,and the subline interval fitting method was proposed based on the trend consistency of the height of the RMS recession curve.This proposed method was tested and validated using the XJTU-SY dataset,and the results of this method were compared with those of the common life prediction methods,such as back propagation neural network(BPNN)and long short-term memory network(LSTM).And the prediction accuracy of the RL of bearings was found to be effectively improved.

关 键 词：故障诊断 滚动轴承 趋势一致性 奇异值替换 子线区间拟合 寿命预测 

分 类 号：TH133.3[机械工程—机械制造及自动化] TH165.3[自动化与计算机技术—检测技术与自动化装置] TP274[自动化与计算机技术—控制科学与工程]