Faint Fault Diagnosis Method for Rolling Bearings Based on Adaptive STWF and Improved OMP
和丹 1魏豪 1胡胜 1王琇峰 2刘晖1
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 西安工程大学 机电工程学院,西安 710048
2. 西安交通大学 机械工程学院, 西安 710049
折叠
摘要
针对工业环境中随机冲击干扰下滚动轴承微弱故障特征提取难题,提出一种基于自适应短时维纳滤波(Adaptive Short Time Wiener Filtering,ASTWF)和改进正交匹配追踪(Orthogonal Matching Pursuit,OMP)的滚动轴承故障特征提取方法.该方法首先采用包络峭度和随余比(Random Shocks and Margin Ratio,RMR)作为联合判据,界定窗长界限并自适应确定STWF最优窗长参数,进而将随机冲击干扰从测试信号中分离出来;然后,利用立方包络自相关谱估计信号中周期频率,构造周期原子库,降低匹配原子冗余度;最后,利用相似性理论优化匹配追踪迭代终止条件,并结合周期原子库,实现弱故障冲击特征快速、准确提取.根据仿真信号和通过变速箱下线检测所得工程数据,可验证所提出方法可有效识别随机冲击干扰下的滚动轴承微弱故障特征.对比最小熵形态反卷积(Minimum Entropy Morphological Deconvolution,MEMD)方法对于随机冲击干扰下滚动轴承微弱故障特征提取效果,发现所提出方法具有更好的故障特征提取能力;与经典OMP方法相比,所提出改进OMP方法信号重构速度提升66%.
Abstract
For the difficulty of faint fault feature extraction of rolling bearings under random shock interference in industrial environment,a new faint fault feature extraction method based on adaptive short time wiener filtering(ASTWF)and improved orthogonal matching pursuit(OMP)is proposed.Firstly,the method uses envelope kurtosis and random shocks-to-margin ratio as joint criteria to define window length bounds and adaptively determine the optimal STWF window length parameters,so as to separate the random shock interference from the test signal.Then,the periodic frequency of the signal is estimated by using the autocorrelation spectrum of the cubic envelope,and the periodic atom library is constructed to reduce the redundancy of the matched atoms.Finally,the similarity theory is used to optimize the matching pursuit iteration termination conditions to realize the accurate and fast extraction of weak fault impact features.According to the simulation signal and the engineering data of gearbox off-line inspection,it is validated that the proposed method can achieve the accurate extraction of weak fault features under random shock interference and effectively diagnose the faint fault of rolling bearings.Compared with the minimum entropy morphological deconvolution(MEMD)method,the proposed method has a better fault feature extraction capability for rolling bearings under random impact interference.Compared with the classical OMP method,the proposed improved OMP method signal reconstruction speed is raised by 66%.
维普
万方数据