基于LLTSA算法维数约简的滚动轴承故障诊断

发布时间：2018-08-12 20:29

【摘要】：针对滚动轴承高维故障特征集识别精度低的问题,提出基于线性局部切空间排列(Linear Local Tangent Space Alignment,LLTSA)算法的维数约简故障诊断模型。首先结合小波包分解、时域、频域及时频域统计方法构造全面表征轴承不同故障特性的混合域特征集,通过敏感度的特征选取方法,从混合特征集中选取轴承故障的敏感特征集,再利用LLTSA算法将高维敏感特征集约简为故障区分度更好的低维特征矢量,并用模糊C均值(Fuzzy C-means,FCM)聚类算法进行故障模式识别,本研究方法能够突出不同特征对分类的贡献率,强化敏感特征,弱化不相关特征,提升了分类精度。最后用深沟球轴承不同部位故障诊断实例验证该模型的有效性。
[Abstract]:Aiming at the problem of low recognition accuracy of high dimensional fault feature sets of rolling bearings, a dimension reduction fault diagnosis model based on linear local tangent space permutation (Linear Local Tangent Space alignment LLTSA) algorithm is proposed. Combining wavelet packet decomposition, time-domain and frequency-domain statistical methods, a hybrid feature set is constructed to fully characterize the different fault characteristics of bearings, and the sensitivity feature selection method is used. The sensitive feature set of bearing fault is selected from the hybrid feature set, and then the high dimension sensitive feature set is reduced to a lower dimension feature vector with better fault differentiation by using LLTSA algorithm, and the fault pattern recognition is carried out by using Fuzzy C-means clustering algorithm. This research method can highlight the contribution rate of different features to classification, strengthen sensitive features, weaken irrelevant features, and improve classification accuracy. Finally, the effectiveness of the model is verified by an example of fault diagnosis in different parts of deep groove ball bearings.
【作者单位】：三峡大学水电机械设备设计与维护湖北省重点实验室;三峡大学机械与动力学院;
【基金】：国家自然科学基金(51205230,51405264) 湖北省自然科学基金(2015CFB445)资助项目
【分类号】：TH133.33

【相似文献】