摘要
针对振动信号的非线性、非平稳特征,提出了一种基于局部保持投影(Locality Preserving Projections,LPP)的转子故障特征提取方法。该方法利用LPP降维后保留数据内部非线性结构的特点,对高维的故障振动信号降维并提取出低维的数据作为特征矢量,采用BP神经网络作为分类器进行故障诊断。实验结果表明,LPP方法能有效提高故障诊断的准确率。
According to the non-linear and non-stationary characteristics of vibration signals,a method of fault feature extraction of rotor based on locality preserving projection（LPP）is proposed.By virtue of its feature that it can project the high dimensional data onto low dimensional data space while maintaining the inherent nonlinear features of structures,the high dimensional vibration signals are reduced to low dimensional space and then is extracted as the characteristic vector.Afterwards,the different faults are classified and diagnosed by the BP neural network.The experiment results show that this method can improve the accuracy of the fault diagnosis.
作者
梁超
路鹏
郜宁
祁伟
LIANG Chao;LU Peng;GAO Ning;QI Wei(Xinjiang Electric Power Research Institute of State Grid Xinjiang Electric Power CO.LTD.,Urumqi 830000,Chin)
出处
《振动工程学报》
EI
CSCD
北大核心
2018年第3期539-544,共6页
Journal of Vibration Engineering
关键词
故障诊断
转子
特征提取
局部保持映射
BP神经网络
fault diagnosis
rotor
feature extraction
locality preserving projections
BP neural network
