基于Schur分解和正交邻域保持嵌入算法的故障数据集降维方法

发布时间：2018-06-11 14:44

本文选题：故障诊断 + 数据降维　；参考：《中国机械工程》2017年21期

【摘要】：针对转子故障特征数据集降维问题,提出一种基于Schur分解和正交邻域保持嵌入算法的故障数据集降维方法——Schur-ONPE降维方法。该方法首先应用小波包分解提取不同频带内的能量以组成故障特征值集合,然后运用Schur分解和ONPE算法将高维特征集向低维投影,使降维后类内散度最小化及类间分离度最大化,最后将降维后得到的低维特征集输入K近邻分类器进行模式识别。通过双跨转子试验台的故障特征数据集进行验证,结果表明该方法能够有效地解决转子故障特征集的降维问题。
[Abstract]:In order to reduce the dimension of rotor fault feature data set, a fault data set reduction method, Schur-ONPE, is proposed based on Schur decomposition and orthogonal neighborhood preserving embedding algorithm. Firstly, the wavelet packet decomposition is used to extract the energy in different frequency bands to form the fault eigenvalue set. Then, Schur decomposition and ONPE algorithm are used to project the high Viterbi set to the lower dimension, which minimizes the intra-class divergence and maximizes the inter-class separation after dimensionality reduction. Finally, the reduced dimension low-Viterbi gather input K-nearest neighbor classifier is used for pattern recognition. The results show that this method can effectively solve the problem of reducing the dimension of rotor fault feature set.
【作者单位】：兰州理工大学机电工程学院;
【基金】：国家自然科学基金资助项目(51675253)
【分类号】：TH17;TP18

【相似文献】