摘要
Abstract
In order to reduce external interference and improve the accuracy of bearing fault diagnosis,this article proposes a new method based on Composite Multiscale Fractional-order Attention Entropy(CMFAE)and Improving Autonomous Learning Multi-model 0-Order(IALMMo-0).First of all,in response to the defects of entropy,this article proposes a Composite Multiscale Fractional-order Attention Entropy(CMFAE),And the original vibration signal is comprehensively extracted using CMFAE.Secondly,in order to avoid more redundant information in the sample,which affects the accuracy rate of fault diagnosis,the Linear Discriminant Analysis(LDA)is used to reduce the dimension of the obtained feature vector.Finally,for the defects of the Autonomous Learning Multi-model 0-Order(ALMMo-0)Classifier,this article proposes to use the information entropy weight method to weight the Mahalanobis distance,and then use Pearson correlation coefficient to improve the weighted Mahalanobis distance,forming Improved Weighted Mahalanobis Distance(IWMD);And IWMD was used to improve its classifier,forming IALMMo-0 classifier;Meanwhile,train the IALMMo-0 classifier using feature vectors and test its performance.In order to test the accuracy and effectiveness of the new methods mentioned in the article,the bearing data was used to analyze in the test,and the accuracy of the failure recognition was as high as 95.601%through test analysis.关键词
轴承/复合多尺度分数阶注意熵/线性判别分析法/零阶自主学习多模型分类器Key words
bearing/composite multiscale fractional-order attention entropy/linear discriminant analysis/au-tonomous learning multi-model 0-order分类
机械工程
