卷积神经网络在机械设备故障诊断领域应用与挑战  

Convolution neural networks for mechanical equipment fault diagnosis: the application and challenge

作　　者：黄鑫 陈仁祥 黄钰[3] HUANG Xin;CHEN Renxiang;HUANG Yu(School of Mechantronics and Vehicle Engineering, Chongqing Jiaotong University, Chongqing 400074, CHN;The State Key Laboratory of Mechanical Transmission, Chongqing University, Chongqing 400030, CHN;School of Automotive and Transportation Engineering, Xihua University, Chengdu 610039,CHN)

机构地区：[1]重庆交通大学机电与车辆工程学院,重庆400074 [2]重庆大学机械传动国家重点实验室,重庆400030 [3]西华大学汽车与交通工程学院,四川成都610039

出　　处：《制造技术与机床》2019年第1期96-100,共5页Manufacturing Technology & Machine Tool

基　　金：重庆市研究生教育创新基金项目资助(CYS18224)

摘　　要：现代机械设备功能及结构日益复杂,其故障呈现出新的特点,使得针对现代机械装备的故障诊断方法遇到了难以克服的技术难题与瓶颈。近年来,卷积神经网络(convolution neural network,CNN)凭借其强大的特征提取与模式识别能力受到学术界和工业界的广泛关注,将CNN应用于机械设备故障诊断的研究已出现端倪。为此,首先介绍CNN实现机械设备故障诊断的原理;然后对CNN实现故障诊断的主要思想和建模方法进行描述;最后总结和讨论了机械设备故障的特点,并讨论CNN在实现对机械装备故障诊断方面所面临的挑战,展望值得继续研究的方向。While the function and structure of modern mechanical equipment become more and more complicated,the fault presents new characteristics,which set a series of challenges and technique problems for fault diagnosis of modern mechanical equipment. In recent years,Convolution Neural Network( CNN) has attracted wide attention from Academia and Industry circle with its powerful feature extraction and pattern recognition ability. The CNN applied in fault diagnosis has emerged. Therefore,the principle of CNN in fault diagnosis is fristly introduced; Then,the main idea and modeling method for CNN in fault diagnosis are depicted; Finally,the characteristics of mechanical equipment failure are summarized and the challenges for CNN in fault diagnosis are discussed as well as the direction that is worth continuing to study is to be expected.

关 键 词：卷积神经网络 机械设备 故障诊断 特征提取与模式识别 

分 类 号：TP227[自动化与计算机技术—检测技术与自动化装置]