基于主元分析与KNN算法的旋转机械故障识别方法

摘要

Considering that it is very difficult to recognize the high-dimensional failure data of rotating machinery System,this paper proposes a novel fault recognition method based on Principle Component Analysis (PCA) and K-nearest neighbour (KNN)algorithm.The time domain and frequency domain characteristic indexes of each state signal are selected in a reasonable way to construct the high dimension feature space.And the high dimension feature space is input to the PCA algorithm to process the dimension reduction thus the low dimensional sensitivity feature is extracted.And then the reduced state samples are input to the KNN algorithm for fault recognition.Simulation results of rolling bearing and rotor illustrate that the method can be used to reduce the feature of high dimension fault samples,and accurately identify the fault samples while realizing the visualization of the sample data Compared with the traditional method,this method has the advantages of simple structure and high recognition rate,and it is of engineering significance to the research of mechanical fault diagnosis.%针对旋转机械高维故障数据难以被准确辨识的情况,提出了一种基于主元分析(principal component analysis,PCA)和K近邻(K-nearest neighbour,KNN)算法的旋转机械故障识别方法.合理选取出各状态信号的时域、频域特征指标构造成高维特征空间,输入给主元分析算法进行降维处理,提取出低维敏感特征,将约简后的状态样本输入给KNN算法进行故障识别.滚动轴承和转子的实验结果表明,该方法能够很好的约简高维故障样本特征,在实现样本数据可视化的同时准确识别出各故障样本.与传统方法相比,该方法具有结构简单、识别率高等优点,对机械故障诊断研究具有一定的工程意义.