基于非平稳时序分析的滚动轴承故障特征提取方法研究

发布时间：2018-08-05 20:06

【摘要】：滚动轴承是旋转机械中使用最为广泛和最易受损的零部件之一,其工作状态直接影响到旋转机械系统的性能,对其进行故障诊断具有重要的实际应用意义。由于滚动轴承运行时产生的振动信号是典型的非平稳随机信号,从振动信号中提取能准确反映轴承运行状态的故障特征是故障诊断的关键。自回归(AR)参数模型是时序分析方法中最基本、实际应用最广的时序模型,但AR模型分析信号是建立在随机平稳性的假设基础上,无法准确分析滚动轴承非平稳随机信号。为此,论文提出一种基于经验模态分解(EMD)和AR模型相结合的滚动轴承故障特征提取方法。该方法用EMD分解将滚动轴承振动信号分解为若干内禀模态函数(IMF),采用相关分析提取前五个IMF分量建立AR模型,提取模型的参数和残差的方差的奇异值作为反映滚动轴承运行状态的特征向量。实验结果表明该方法提取特征精度较高,效果较好。为克服EMD分解信号精度不高的问题,论文利用小波包分解(WPD)具有良好的时频局部化特性和多分辨率的特征,将非平稳振动信号转化为平稳信号的基础上,把振动信号分解到各个频段中使AR模型能有效跟踪信号,提出了基于小波包自回归(WPD_AR)参数模型和小波包时变自回归(WPD_TVAR)参数模型的两种滚动轴承故障特征提取方法。首先对滚动轴承振动信号进行小波包分解,然后对分解得到的各结点系数分别建立AR模型和TVAR模型,分别提取WPD_AR模型和WPD_TVAR模型的参数奇异值作为反映滚动轴承运行状态的特征向量。实验结果表明,WPD_AR模型比EMD_AR模型故障特征提取的效果更明显,速度更快；WPD_TVAR模型比WPD_AR模型故障特征提取的结果更准确,精度更高。将提出的三种故障特征提取方法获取的故障特征送入支持向量机分类器进行故障分类与诊断。实验表明,本文所提方法可以有效、准确地识别滚动轴承的故障状况,验证了论文提出的基于非平稳时序分析的滚动轴承故障特征提取方法的有效性。
[Abstract]:Rolling bearing is one of the most widely used and easily damaged parts in rotating machinery. Its working state directly affects the performance of rotating machinery system. Because the vibration signal generated by rolling bearing is a typical non-stationary random signal, it is the key to fault diagnosis to extract the fault characteristics from the vibration signal which can accurately reflect the running state of the bearing. The autoregressive (AR) parameter model is the most basic and widely used time series model in time series analysis, but the AR model is based on the assumption of stochastic stationarity and can not accurately analyze the non-stationary random signals of rolling bearings. In this paper, a fault feature extraction method based on empirical mode decomposition (EMD) and AR model is proposed. In this method, the vibration signal of rolling bearing is decomposed into several intrinsic modal functions by EMD decomposition. The first five IMF components are extracted by correlation analysis to establish AR model. The singular values of the parameters of the model and the variance of the residuals are extracted as the eigenvectors to reflect the running state of the rolling bearings. The experimental results show that this method has high accuracy and good effect. In order to overcome the problem that the precision of EMD decomposition signal is not high, this paper uses wavelet packet decomposition (WPD) to have good time-frequency localization and multi-resolution characteristics, and converts the non-stationary vibration signal into stationary signal. Based on wavelet packet autoregressive (WPD_AR) parameter model and wavelet packet time-varying autoregressive (WPD_TVAR) parameter model, two fault feature extraction methods for rolling bearing are proposed. Firstly, the vibration signal of rolling bearing is decomposed by wavelet packet, then AR model and TVAR model are established for each node coefficient obtained by decomposition. The singular values of the WPD_AR model and the WPD_TVAR model are extracted as the characteristic vectors to reflect the running state of the rolling bearing. The experimental results show that the effect of fault feature extraction by WPD-AR model is more obvious than that by EMD_AR model, and the speed of fault feature extraction by WPD-VAR model is more accurate and the accuracy is higher than that by WPD_AR model. The fault features obtained by the three fault feature extraction methods are sent to the support vector machine classifier for fault classification and diagnosis. Experiments show that the proposed method can effectively and accurately identify the fault status of rolling bearings and verify the effectiveness of the fault feature extraction method proposed in this paper based on non-stationary time series analysis.
【学位授予单位】：中南大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】