基于改进阈值DT-CWT降噪与WPHM模型的滚动轴承可靠度评估  

Reliability evaluation of rolling bearing based on improved threshold DT-CWT noise reduction and WPHM model

作　　者：赵红美[1] 周志宏 杨胜兵 ZHAO Hong-mei;ZHOU Zhi-hong;YANG Sheng-bing(College of Mechanical Engineering,Tangshan Polytechnic College,Tangshan 063299,China;College of Mechanical Engineering and Automation,Fuzhou University,Fuzhou 350108,China;Fujian Yongan Bearing Co.,Ltd.,Sanming 366000,China)

机构地区：[1]唐山工业职业技术学院机械工程学院,河北唐山063299 [2]福州大学机械工程及自动化学院,福建福州350108 [3]福建省永安轴承有限公司,福建三明366000

出　　处：《机电工程》2022年第9期1235-1242,共8页Journal of Mechanical & Electrical Engineering

基　　金：福建省自然科学基金资助项目(2017J01691)。

摘　　要：针对滚动轴承运行数据集降噪难度大,且其可靠度评估模型预测精度不高的问题,提出了一种基于改进阈值DT-CWT降噪与WPHM模型的滚动轴承可靠度评估方法。首先,在数据降噪方面,利用一种改进的双树离散连续小波变换(DT-CWT)阈值降噪方法,将尺度因子和平移因子离散化,通过两组平行且独立的低通和高通滤波器,构成实部树和虚部树,实现了对信号的完全重构;然后,在数据处理方面,采用了PRONOSTIA实验台的全寿命实验数据,把粒子群优化(PSO)的全局最优搜索策略与最小二乘(LS)进行了融合,得到了威布尔比例风险模型(WPHM)的最佳参数;最后,采用法国弗朗什孔泰大学FEMTO的PRONOSTIA轴承实验台数据,对改进DT-CWT阈值降噪效果进行了评估,并对通过WPHM模型计算得到的可靠度评估曲线进行了验证。研究结果表明:(1)相对于传统阈值函数,改进阈值DT-CWT函数降噪效果更好,可提高信噪比(SNR)58.2%,降低均方根差(RMSE)58.3%,降噪信号变化趋势与原信号保持一致;(2)利用全寿命实验数据和PSO-LS方法对β、η和γ进行了参数估计,解决了WPHM模型中的参数估计问题;(3)可靠度评估曲线与轴承实际退化状态相符,WPHM模型可以反映轴承的健康状态。Aiming at the difficulty of noise reduction of rolling bearing operation data set and the low prediction accuracy of its reliability evaluation model,a reliability evaluation method of rolling bearing based on improved threshold DT-CWT noise reduction and WPHM model was proposed.Firstly,in data noise reduction,the dual-tree continuous wavelet transform(DT-CWT)was used to discretize the scale factor and translation factor,and the signal was completely reconstructed by two sets of parallel and independent low-pass and high-pass filters to form a real part tree and an imaginary part tree.Secondly,in terms of data processing,the global optimal search strategy of particle swarm optimization(PSO)was fused with least squares(LS)to quickly find the optimal parameters of Weibull proportional hazards model(WPHM)using the full-life test data of PRONOSTIA experimental bench.Finally,using the data of PRONOSTIA bearing test-bed of FEMTO,FRANCESCONTE University,France,the noise reduction effect of the improved DT-CWT threshold was evaluated,and the reliability evaluation curve calculated by WPHM model was verified.The results indicate that,compared with the traditional threshold function,the improved threshold DT-CWT function has better noise reduction effect,the signal-to-noise ratio(SNR)can be improved to 58.2% and the root mean square error(RMSE)can be reduced to 58.3% by the DT-CWT,and the trend of the noise reduction signal remains the same as the original signal.Parameter estimation of β,η and γ using full-life experimental data and PSO-LS method,solving the parameter estimation problem in WPHM model.The reliability evaluation curve is consistent with the actual degradation state of the bearing,and the WPHM model can reflect the health state of the bearing.

关 键 词：轴承性能退化 双树离散连续小波变换 粒子群优化 最小二乘算法 威布尔比例风险模型 数据降噪算法 最佳参数 

分 类 号：TH133.33[机械工程—机械制造及自动化]