卷积神经网络和峭度在轴承故障诊断中的应用

针对传统智能诊断方法依靠专家知识和人工提取数据特征工作量大的问题,结合深度学习方法在特征提取和处理大数据方面的优势,研究了一种基于卷积神经网络和振动信号峭度指标的滚动轴承故障诊断方法。该方法将深度学习应用于轴承故障诊断,提取滚动轴承正常状态、内圈故障、外圈故障和滚动体故障4种状态的振动信号,将振动信号分段处理得到峭度指标,使用数据到图像的转换方法将峭度指标转换为灰度图,送入卷积神经网络模型完成故障分类。在进行滚动轴承故障诊断的实验时,所提的模型诊断准确率达到99.5%,高于传统支持向量机(SVM)算法的95.8%。 

Application of convolutional neural network and kurtosis in fault diagnosis of rolling bearing

Traditional intelligent diagnosis method relying much on expert knowledge and manual extraction data features takes a lot of work. Based on the advantages of deep learning in feature extraction and processing of big data,a method of rolling bearing fault diagnosis based on convolution neural network and kurtosis was studied. This method was used to analyse four kinds of vibration signal of the normal state,the inner race fault,the outer race fault and the ball fault. The vibration signal was processed in segments to obtain kurtosis, which was converted into gray images by data-to-image method. Finally, these were fed into convolution neural network model to complete rolling bearing fault classification. In the case of rolling bearing fault diagnosis, the improved model had a diagnostic accuracy of 99.5%, which was higher than 95.8% of the traditional support vector machine (SVM) algorithm.