摘要
针对轴承复合故障振动信号的多分量耦合调制特征及特征参数难确定问题,提出一种基于深度学习的城轨列车轴承复合故障诊断方法。对轴承振动信号进行标准化处理并转换为二维数组,将数组以灰度图形式存储得到特征样本,分为训练集和测试集。将训练集作为卷积神经网络(CNN)模型的输入,对模型进行训练,确定网络最佳结构和参数;通过测试集验证网络的可行性和有效性。实验结果表明,基于深度学习的城轨列车轴承复合故障诊断方法,可有效识别城轨列车轴承复合故障,为轴承复合故障辨识提供了一种新思路。
To address the difficulty in determining the multi-component coupling modulation characteristics and characteristic parameters of the vibration signals in bearing compound faults,a compound fault diagnosis method for bearings of urban rail train based on deep learning was proposed.Firstly,in order to obtain the feature samples,the vibration signal of the bearing was standardized and transferred into two-dimensional array,which was saved in the form of grayscale.Then the feature samples were divided into training set and testing set.Secondly,the training set was used as the input of the Convolutional Neural Network(CNN)model,to train the model and determine the optimal structure and parameters of the network.Finally,the feasibility and validity of the network were verified by means of the test set.The research results show that this method can effectively recognize the compound faults of the bearings of urban rail trains and provide a new thinking to identify such faults.
作者
姚德臣
刘恒畅
杨建伟
李熙
崔晓飞
YAO Dechen;LIU Hengchang;YANG Jianwei;LI Xi;CUI Xiaofei(School of Mechanical-Electronic and Vehicle Engineering,Beijing University of Civil Engineering Architecture,Beijing 100044,China;Beijing Key Laboratory of Performance Guarantee on Urban Rail Transit Vehicles,Beijing University of Civil Engineering Architecture,Beijing 100044,China;Beijing Mass Transit Railway Operation Co.,Ltd.,Beijing 100044,China)
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2021年第6期37-44,共8页
Journal of the China Railway Society
基金
国家自然科学基金(51605023)
北京市百千万人才工程(2014018)
国家重点研发计划(2016YFB1200402)
北京市教委科研计划(SQKM201810016015)
北京建筑大学研究生创新项目(PG2019092)
建大英才培养计划和北京建筑大学科学研究基金(00331615015)。
关键词
城轨列车轴承
复合故障
卷积神经网络
故障诊断
urban rail train bearing
compound fault
convolution neural network
fault diagnosis
作者简介
姚德臣(1984-),男,山东德州人,副教授,博士。E-mail:yaodechen@bucea.edu.cn。
