深度学习在故障诊断中的应用综述  

Application Review of Deep Learning in Fault Diagnosis

作　　者：李春林 熊建斌 苏乃权 张清华 梁琼 叶宝玉[5] LI Chunlin;XIONG Jianbin;SU Naiquan;ZHANG Qinghua;LIANG Qiong;YE Baoyu(School of Automation, Guangdong Polytechnic Normal University, Guangzhou Guangdong 510665, China;College of Mechanical and Electrical Engineering, Guangdong University of Technology, Guangzhou Guangdong 510006, China;Guangdong Provincial Key Lab of Fault Diagnosis of Petrochemical Equipment, Maoming Guangdong 525000, China;School of Computing, Guangdong Polytechnic Normal University,Guangzhou Guangdong 510665,China;Guangzhou Civil Aviation College, Guangzhou Guangdong 510403, China)

机构地区：[1]广东技术师范大学自动化学院,广东广州510665 [2]广东工业大学机电学院,广东广州510006 [3]广东省石化装备故障诊断重点实验室,广东茂名525000 [4]广东技术师范大学计算机学院,广东广州510665 [5]广州民航职业技术学院,广东广州510403

出　　处：《机床与液压》2020年第13期174-184,共11页Machine Tool & Hydraulics

基　　金：国家自然科学基金面上项目(61473331);广东技术师范学院人才引进项目(991512203,991560236);广东省自然科学基金项目(2019A1515010700,2018A030307038);广东省科技重大专项项目(2017B030305004);广东省普通高校重点项目(2019KZDXM020,2019KZDZX1004,2017KZXM052);广州市科技计划项目(201903010059)。

摘　　要：阐述了深度学习在故障诊断和图像分析、语音识别和文本理解等领域的应用;介绍卷积神经网络、深度置信网络、堆叠自动编码网络、递归神经网络4种典型的深度学习模型;综述近几年深度学习在故障诊断中的模型选择、学习算法和实际应用等方面的研究新进;探讨深度学习在故障诊断中的理论分析、特征提取、优化训练和研究拓展等。The applications of deep learning in fault diagnosis,image analysis,speech recognition and text understanding were described.Four typical deep learning models of convolutional neural network,deep confidence network,stack automatic coding network and recursive neural network were introduced.The research progress of deep learning in model selection,learning algorithm and practical application in fault diagnosis were reviewed.The theoretical analysis,feature extraction,optimization training and research development of deep learning in fault diagnosis were discussed.

关 键 词：故障诊断 深度学习 特征识别 神经网络 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]