正交局部保持投影早期故障特征提取方法

Feature Extraction for Incipient Diagnosis of Rolling Bearings Based on Orthogonal Locality Preserving Projection

针对轴承早期故障特征难以提取,提出了一种基于正交局部保持投影的轴承故障特征提取方法。由时域指标和小波频带能量组成高维特征空间。运用正交局部保持投影方法通过训练样本数据求出正交转换矩阵,测试样本经正交转换矩阵转化后得到低维向量。利用不同故障样本的类间散度和同种故障样本的类内散度两个指标来衡量该方法的有效性,通过滚动轴承故障数据的仿真,证明提出的正交局部保持投影的特征提取方法是有效的。

A feature extraction method based on orthogonal locality preserving projection algorithm for incipient diagnosis of rolling bearings is put forward because of the difficulty in extracting the incipient diagnosis features of rolling bearings. Constructing the high- dimensional feature space with the time domain indexes and the wavelet frequency domain energies. the algorithm is employed to obtain low- dimensional orthogonal transformation matrix by training samples,Low- dimensional vectors are produced,which have been transformed from tested samples by low-dimensional orthogonal transformation matrix 。We choose two indexes as a measure to judge the performance of the proposed method： the divergence between the class,the divergence within the class. Through the simulation experiment,the fault feature extraction method for rolling bearing is effective.