基于机器学习理论的机械故障诊断方法综述  

Review on mechanical fault diagnosis methods based on machine learning theories

作　　者：肖乾浩 XIAO Qianhao(College of Electromechanical Engineering,Qingdao University of Science and Technology,Qingdao 266061,China)

机构地区：[1]青岛科技大学机电工程学院,青岛266061

出　　处：《现代制造工程》2021年第7期148-161,共14页Modern Manufacturing Engineering

摘　　要：现代机械设备日趋精密化、智能化,同时工作环境与工况也越来越复杂,一旦零/部件出现故障,不但会对设备本身造成伤害,还有可能造成人员伤亡等事故;因此,及时有效地发现并处理设备故障有着重要的意义。随着人工智能技术的发展,以机器学习为核心的机械故障诊断技术飞速发展。对常用的机器学习理论进行了梳理和总结,主要介绍了基于浅层学习下的人工神经网络、支持向量机及Boosting算法和基于深度学习下的卷积神经网络、自动编码器及深度置信网络这6大类机器学习模型,分析比较了这些模型的优缺点,并总结了各个模型在机械故障诊断领域的应用。最后对机器学习在故障诊断领域未来的发展做出了总结和展望。Modern mechanical equipment is becoming more and more sophisticated and intelligent,and the working environment and working conditions are becoming more and more complex.Once the failure of parts and components occurs,it will not only cause damage to the equipment itself,but also cause casualties and other problems.Therefore,it is of great significance to find and deal with equipment faults timely and effectively.With the development of artificial intelligence technology,the mechanical fault diagnosis technology with machine learning as the core has developed rapidly.The commonly used machine learning theories are combed and summarized,and six kinds of machine learning models are mainly introduced:artificial neural network,support vector machine,Boosting algorithm based on shallow learning,and convolutional neural network,automatic encoder,deep belief network based on deep learning.The advantages and disadvantages of these models are analyzed and compared,and the applications of these models in the field of mechanical fault diagnosis are summarized.Finally,the future development of machine learning in the field of fault diagnosis is summarized and prospected.

关 键 词：机器学习 故障诊断 浅层学习 深度学习 

分 类 号：TH117.2[机械工程—机械设计及理论]