作者：杨劼立1, 林建辉1, 谌亮2
作者单位：1. 西南交通大学机械工程学院，四川 成都 610031;
2. 中车长春轨道客车股份有限公司，吉林 长春 130000
关键词：列车;轴承;故障诊断;MED;CNN
摘要：
为增强基于振动信号的列车滚动轴承故障的诊断准确性，提出一种采用MED辅助特征提取的卷积神经网络模型。首先采用MED理论对振动信号进行处理，再将其与原信号构成二维张量送入卷积神经网络进行训练。这样，既在一定程度上突出信号中故障引起的冲击成分，使得故障特征更容易被卷积神经网络提取出来，也完整地保留原信号中的信息，不影响信息的完整性。采用实测轴承数据进行性能分析和验证，对比直接使用CNN的方法。结果表明：该模型确拥有更好的性能，在测试集与训练集来自于不同运行速度数据的情况下，表现出更好的泛化能力，更高的诊断准确性，将测试集的诊断准确率提高2个百分点，是一种能更好用于列车滚动轴承故障智能诊断的方法。
Fault diagnosis method for train bearings based on MED-assisted feature extraction CNN model
YANG Jieli1, LIN Jianhui1, CHEN Liang2
1. College of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China;
2. CRRC Changchun Railway Vehicles Co., Ltd., Changchun 130000, China
Abstract: In order to enhance the diagnostic accuracy of rolling bearing faults of trains based on vibration signals, a CNN model with MED-assisted feature extraction is proposed. First, use MED theory to process the vibration signal, and then form a two-dimensional tensor with the original signal and send it to the convolutional neural network for training. In this way, the shock component caused by the fault in the signal is highlighted to a certain extent, so that the fault features are more easily extracted by the convolutional neural network, and the information in the original signal is completely retained without affecting the information integrity. Performance analysis and verification were performed using measured bearing data, and the method of using CNN directly was compared. The results show that the model does have better performance: better generalization ability, and higher diagnostic accuracy. It is a method that can be applied to the intelligent diagnosis of rolling bearing faults in trains.
Keywords: train;bearing;fault diagnosis;MED;CNN
2020, 46(10):124-129 收稿日期: 2020-02-09;收到修改稿日期: 2020-04-10
基金项目: 国家重点研发计划（2018YFB1201904-04）
作者简介: 杨劼立(1994-),男,四川成都市人,硕士研究生,专业方向为机械故障诊断
参考文献
[1] MOHANTY S, GUPTA K K, RAJU K S. Hurst based vibro-acoustic feature extraction of bearing using EMD and VMD[J]. Measurement, 2018, 117: 200-220
[2] 胡爱军, 赵军, 孙尚飞, 等. 基于相关峭度共振解调的滚动轴承复合故障特征分离方法[J]. 振动与冲击, 2019, 38(8): 110-116
[3] 张琛, 赵荣珍, 邓林峰. 基于EEMD奇异值熵的滚动轴承故障诊断方法[J]. 振动.测试与诊断, 2019, 39(2): 353-358,446-447
[4] SHAO H, JIANG H, ZHANG X, et al. Rolling bearing fault diagnosis using an optimization deep belief network[J]. Measurement Science and Technology, 2015, 26(11): 115002
[5] 陈仁祥, 黄鑫, 杨黎霞, 等. 基于卷积神经网络和离散小波变换的滚动轴承故障诊断[J]. 振动工程学报, 2018, 31(5): 883-891
[6] 杨平, 苏燕辰. 基于卷积门控循环网络的滚动轴承故障诊断[J]. 航空动力学报, 2019, 34(11): 2432-2439
[7] QIAO H, WANG T, WANG P, et al. An adaptive weighted multiscale convolutional neural network for rotating machinery fault diagnosis under variable operating conditions[J]. IEEE Access, 2019, 7: 118954-118964
[8] WIGGINS R A. Minimum entropy deconvolution[J]. Geoexploration, 1978, 9(16): 21-35
[9] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks [C]//International Conference on Neural Information Processing Systems. 2012
[10] BOUVRIE J. Notes on convolutional neural networks[R]. Center for Biological and Computational Learning, 2006.
[11] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324
