基于ELDA降维与MPA-SVM的滚动轴承故障诊断方法
Rolling Bearing Fault Diagnosis Method Based on ELDA Dimension Reduction and MPA-SVM
刘运航 1宋宇博 1朱大鹏2
作者信息
- 1. 兰州交通大学 机电技术研究所,兰州 730070
- 2. 兰州交通大学 交通运输学院,兰州 730070
- 折叠
摘要
为了提高滚动轴承故障诊断精度,提出一种基于偏心线性判别分析(Eccentric Linear Discriminant Analysis,ELDA)降维算法与经海洋捕食者算法(Marine Predators Algorithm,MPA)优化的支持向量机(Support Vector Machine,SVM)相结合的滚动轴承故障诊断方法.首先对轴承信号应用时域和频域分析方法构建高维特征集,其次应用自适应最大似然估计方法(Adaptive Maximum Likelihood Estimation,AMLE)进行固有维度估计,利用ELDA算法进行二次特征提取,充分挖掘敏感特征,降低冗余特征对故障诊断的影响;最后将低维敏感可分矩阵输入到MPA-SVM分类器中识别故障类型.实验分析表明,所提方法能有效缩短训练时长并提高诊断准确率.
Abstract
In order to raise the fault diagnosis accuracy of rolling bearings,a new rolling bearing fault classification method based on Eccentric Linear Discriminant Analysis(ELDA)dimension reduction algorithm and Marine Predators Algorithm(MPA)optimized support vector machine was proposed.Firstly,the time domain and frequency domain analysis methods were used to construct the high dimension eigenmatrix of bearing signals.Then,the Adaptive Maximum Likelihood Estimation(AMLE)method was used to estimate the intrinsic dimension.The ELDA algorithm was used to extract the secondary features,so as to fully explore the sensitive features and reduce the impact of redundant features on fault diagnosis.Finally,the low-dimensional sensitive separable matrix was input into the MPA-SVM classifier to identify the fault type.Experimental analysis shows that the proposed method can effectively shorten the training time and improve the accuracy of diagnosis.
关键词
故障诊断/滚动轴承/特征降维/海洋捕食者算法/支持向量机Key words
fault diagnosis/rolling bearing/feature dimension reduction/marine predator algorithm/support vector machine引用本文复制引用
基金项目
甘肃省教育厅青年博士基金(2021QB-053)
国家自然科学基金(51765028)
出版年
2024
国家科技期刊平台
NSTL
万方数据