摘要
工业设备数据通常具有“多变量+时序”的复杂二维特性,而在当前工业设备故障诊断领域,主流的特征提取方法只提取了一维特征。针对特征提取不充分的问题,提出一种基于2DSVD特征提取的工业设备故障诊断方法,该方法首先使用故障诊断训练样本集构造行-行以及列-列的协方差矩阵,然后通过协方差矩阵的特征向量提取样本特征,最后结合KNN算法对特征集合进行分类,实现故障诊断。使用2DSVD方法提取的特征矩阵不仅行数及列数比原故障诊断样本低,而且充分考虑了样本的二维特性。同时,构建了基于2DSVD+KNN的故障诊断模型,并利用自动洗车机实验数据与传统的只基于1维特征提取的方法以及无特征提取的方法进行对比实验,实验结果表明基于2DSVD特征提取方法的故障诊断准确率有明显的提高。
The data which generated from industrial equipment usually has the complex two-dimensional characteristics of"multivariable+time series".However,in the field of industrial equipment fault diagnosis,only one-dimensional feature was extracted by the mainstream feature extraction methods.In order to solve the problem of inadequate for feature extraction,the industrial equipment fault diagnosis method based on 2DSVD of feature extraction is proposed.First of all,the training sample set of fault diagnosis is used to structure row-row and column-column covariance matrixes,and then the sample characteristics are extracted by the characteristic vector of covariance matrixes.Finally,combining classifying feature collection KNN algorithm,realize fault diagnosis.The number of rows and columns extracted by 2DSVD method is not only lower than that of the original fault diagnosis samples,but also the two-dimensional characteristics of the samples are fully considered.At the same time,the fault diagnosis model based on 2DSVD+KNN was constructed,and the experimental data of automatic car washing machine were compared with the traditional method based only on 1-dimensional feature extraction and the method without feature extraction.The experimental results show that the fault diagnosis accuracy based on 2DSVD feature extraction method has been significantly improved.
作者
王正家
刘鸣
何嘉奇
陈长乐
WANG Zheng-jia;LIU Ming;HE Jia-qi;CHEN Chang-le(School of Mechanical Engineering,Hubei University of Technology,Wuhan 430068,China;Hubei Key Laboratory of Modern Manufacture Quality Engineering,Hubei University of Technology,Wuhan 430068,China)
出处
《组合机床与自动化加工技术》
北大核心
2021年第12期49-52,57,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金项目(51575164)。
