基于小波域对数正态模型的滚动轴承故障诊断

摘要

针对小波分析无法全面准确描述滚动轴承振动信号的非高斯问题,提出一种结合小波变换与对数正态分布模型的故障特征提取方法,以提取能准确反映滚动轴承运行状态的特征信息。首先,通过小波变换对滚动轴承运行时产生的非平稳、非高斯振动信号进行分解重构,得到不同尺度下的重构信号;然后对重构信号建立对数正态分布模型,提取模型的对数均值和对数标准差作为表征滚动轴承运行状态的统计特征;最后采用支持向量机分类器对提取的特征进行故障分类与识别。实验结果表明,该方法可以有效、准确地识别滚动轴承的运行状态。%Since the non-Gaussian of rolling bearing vibration signals can not be fully described by the wavelet analysis,a feature extraction approach based on wavelet transformation and lognormal model was proposed,so as to that the feature vectors were extracted to reflect accurately the running state of rolling bearings.First of all,the non-stationary and non-Gaussion signals generated by rolling bearing vibrations were decomposed into some coefficients by wavelet transformation.Then the signals of single reconstruction were modeled as lognormal models and its log mean and log variance were extracted.Finally,fault patterns were recognized by the feature vectors using support vector machine（SVM） classifier.The experimental results show the effectiveness and accuracy of the proposed approach for recognizing the states of rolling bearings.