文献类型：期刊文献

中文题名：基于深度CNN和ELM的滚动轴承故障诊断研究

英文题名：Research on Fault Diagnosis of Rolling Bearings Based on Deep CNN and ELM

作者：顾鑫 唐向红 陆见光 黎书文

第一作者：顾鑫

机构：[1]贵州大学现代制造技术教育部重点实验室,贵阳550025;[2]贵州大学机械工程学院,贵阳550025;[3]贵州大学公共大数据国家重点实验室,贵阳550025;[4]贵州理工学院机械工程学院,贵阳550003

第一机构：贵州大学现代制造技术教育部重点实验室,贵阳550025

年份：2020

卷号：41

期号：3

起止页码：154-158

中文期刊名：兵器装备工程学报

外文期刊名：Journal of Ordnance Equipment Engineering

收录：CSTPCD;;北大核心:【北大核心2017】；

基金：贵州省公共大数据重点实验室开放基金项目(2017BDKFJJ019);贵州省科技计划项目(黔科合平台人才[2017]5789-10);贵州省科技计划项目(黔科合平台人才[2017]5789-24)。

语种：中文

中文关键词：深度卷积神经网络;极限学习机;滚动轴承;故障诊断;实时性

外文关键词：deep convolutional neural network;extreme learning machine;rolling bearings;fault diagnosis;real-time performance

摘要：提出了一种深度卷积神经网络与极限学习机相结合的滚动轴承自适应故障诊断方法。该方法的第一阶段训练深度卷积神经网络作为特征提取器:通过卷积层和池化层提取低阶特征,然后在全连接层合成高层次特征。第二阶段将第一阶段自适应提取出来的特征通过极限学习机进行轴承故障类别的准确快速分类,实现了自适应“端到端”的故障诊断。实验结果表明,该方法能有效的识别故障类别,缩短了训练时间,并具有良好的鲁棒性和实时性。
A method for rolling bearing fault diagnosis combined with deep convolutional neural network(DCNN)and extreme learning machine(ELM)was proposed.The method was divided into two steps:the deep convolutional neural network was trained as a feature extractor:extracted low-level features through the convolutional layer and the pooling layer,and then merged high-level features in the fully connected layer;the extracted features were classified by the extreme learning machine to achieve an adaptive“end-to-end”fault diagnosis.The experimental results show that the method can identify fault categories effectively,shorten training time,and has good robustness and real-time performance.