[1]刘崇睿,缪炳荣,赵龙,等.基于改进DCGAN和VGG16的小样本车轮踏面损伤识别模型[J].高速铁路技术,2026,(01):63-70.[doi:10.12098/j.issn.1674-8247.2026.01.010]
　LIU Chongrui,MIAO Bingrong,ZHAO Long,et al.Small Sample Wheel Tread Damage Recognition Model Based on Improved DCGAN and VGG16[J].HIGH SPEED RAILWAY TECHNOLOGY,2026,(01):63-70.[doi:10.12098/j.issn.1674-8247.2026.01.010]
点击复制

基于改进DCGAN和VGG16的小样本车轮踏面损伤识别模型()

《高速铁路技术》[ISSN:1674-8247/CN:51-1730/U]

卷:
期数:
2026年01期
页码:
63-70
栏目:
研究创新
出版日期:
2026-01-30

文章信息/Info

Title:
Small Sample Wheel Tread Damage Recognition Model Based on Improved DCGAN and VGG16
文章编号:
1674-8247(2026)01-0063-08
作者:
刘崇睿1; 缪炳荣1; 赵龙2; 徐松源1; 金月皓1
(1. 西南交通大学轨道交通运载系统全国重点实验室, 成都 610031; 2. 中铁二院工程集团有限责任公司, 成都610031)
Author(s):
LIU Chongrui1;  MIAO Bingrong1;  ZHAO Long2;  XU Songyuan1;  JIN Yuehao1
(1.State Key Laboratory of Rail Transit Vehicle System, Southwest Jiaotong University, Chengdu 610031, China; 2. China Railway Eryuan Engineering Group Co., Ltd., Chengdu 610031, China)
关键词:
深度卷积生成对抗网络;  小样本;  机器学习;  车轮踏面损伤;  小波变换
Keywords:
Deep Convolutional Generative Adversarial Network(DCGAN);  small sample;  machine learning;  wheel tread damage;  wavelet transform
分类号:
U279.3+23
DOI:
10.12098/j.issn.1674-8247.2026.01.010
文献标志码:
A
摘要:
目前在复杂行车环境下,获取大量真实车轮故障样本数据面临诸多困难,导致训练数据呈小样本特征。为提高在小样本数据下车轮踏面损伤识别的精度与效率,本文提出一种融合通道注意力机制、空洞卷积的改进深度卷积生成对抗网络(DCGAN)与VGG16模型相结合的车轮踏面损伤识别方法。首先,搭建车辆-轨道刚柔耦合动力学模型并采集车轮在不同损伤下轴箱振动加速度信号; 其次,通过Morlet小波变换提取振动信号的二维时频特征; 随后,利用改进的DCGAN进行训练集数据扩充; 最后,借助VGG16分类模型对车轮踏面损伤程度进行分类。结果表明,采用本文所提出的改进DCGAN对训练集进行扩充后,识别准确率达98.36%,所提方法对车轮踏面损伤具有良好的识别效果。
Abstract:
Currently, acquiring substantial real wheel fault sample data under complex operating environments presents numerous challenges, resulting in training data exhibiting small-sample characteristics. To improve the accuracy and efficiency of wheel tread damage recognition under small-sample data conditions, this paper proposed a wheel tread damage identification method that integrates an improved deep convolutional generative adversarial network(DCGAN)incorporating a channel attention mechanism and dilated convolutions, combined with the VGG16 model. First, a vehicle-track rigid-flexible coupled dynamics model was established to collect axle box vibration acceleration signals under various wheel damage conditions. Then, two-dimensional time-frequency features of the vibration signals were extracted using Morlet wavelet transform. Next, the improved DCGAN was used to augment the training dataset. Finally, the VGG16 classification model was employed to classify the severity of wheel tread damage. The results show that after augmenting the training dataset using the proposed improved DCGAN, the recognition accuracy reaches 98.36%, indicating that the proposed method has good performance in identifying wheel tread damage.

参考文献/References:

[1] 敬霖, 刘凯. 车轮踏面缺陷引起的轮轨动态响应综述 [J]. 交通运输工程学报, 2021, 21(1): 285-315.
JING Lin, LIU Kai. Review on Wheel-rail Dynamic Responses Caused by Wheel Tread Defects[J]. Journal of Traffic and Transportation Engineering, 2021, 21(1): 285-315.
[2] 曾京, 彭莘宇, 汪群生, 等. 铁道车辆车轮扁疤故障检测技术综述[J]. 交通运输工程学报, 2022, 22(2): 1-18.
ZENG Jing, PENG Xinyu, WANG Qunsheng, et al. Review on Detection Technologies of Railway Vehicle Wheel Flat Fault [J]. Journal of Traffic and Transportation Engineering, 2022, 22(2): 1-18.
[3] 缪炳荣, 刘俊利, 张盈, 等. 轨道车辆结构振动损伤识别技术综述[J]. 交通运输工程学报, 2021, 21(1): 338-357.
MIAO Bingrong, LIU Junli, ZHANG Ying, et al. Review on Structural Vibration Damage Identification Technology for Railway Vehicles[J]. Journal of Traffic and Transportation Engineering, 2021, 21(1): 338-357.
[4] 谢利明, 高晓蓉, 罗林, 等. 车轮踏面不圆度在线监测技术的现状与分析[J]. 铁道技术监督, 2012, 40(6): 12-14.
XIE Liming, GAO Xiaorong, LUO Lin, et al. Current Status and Analysis of Wheel Tread Out-of-Roundness On-line Monitoring Technology[J]. Railway Quality Control, 2012, 40(6): 12-14.
[5] SRESAKOOLCHAI J, KAEWUNRUEN S.Wheel Flat Detection and Severity Classification Using Deep Learning Techniques[J]. Insight: Non-destructive Testing and Condition Monitoring, 2021(7): 63.
[6] SHAIKH K, HUSSAIN I, CHOWDHRY B S. Wheel Defect Detection Using a Hybrid Deep Learning Approach[J]. Sensors, 2023, 23(14): 6248.
[7] 李恒, 张氢, 秦仙蓉, 等. 基于短时傅里叶变换和卷积神经网络的轴承故障诊断方法[J]. 振动与冲击, 2018, 37(19): 124-131.
LI Heng, ZHANG Qing, QIN Xianrong, et al. Fault Diagnosis Method for Rolling Bearings Based on Short-time Fourier Transform and Convolution Neural Network[J]. Journal of Vibration and Shock, 2018, 37(19): 124-131.
[8] 张安安, 黄晋英, 冀树伟, 等. 基于卷积神经网络图像分类的轴承故障模式识别[J]. 振动与冲击, 2020, 39(4): 165-171.
ZHANG An'an, HUANG Jinying, JI Shuwei, et al. Bearing Fault Pattern Recognition Based on Image Classification with CNN[J]. Journal of Vibration and Shock, 2020, 39(4): 165-171.
[9] 李大柱, 牛江, 梁树林, 等. 基于多尺度时频图与卷积神经网络的车轮故障智能诊断[J]. 铁道科学与工程学报, 2023, 20(3): 1032-1043.
LI Dazhu, NIU Jiang, LIANG Shulin, et al. Intelligent Wheel Fault Diagnosis Based on Multi-scale Time-frequency Map and Convolutional Neural Network[J]. Journal of Railway Science and Engineering, 2023, 20(3): 1032-1043.
[10] 李凌轩, 马振玮, 于泽峻, 等. 基于MTF和改进残差网络的轴承故障定量诊断方法[J]. 东北大学学报(自然科学版), 2024, 45(5): 697-706.
LI Lingxuan, MA Zhenwei, YU Zejun, et al. Method for Bearing Fault Quantitative Diagnosis Based on MTF and Improved Residual Network[J]. Journal of Northeastern University(Natural Science), 2024, 45(5): 697-706.
[11] LI Yanting, JIANG Wenbo, ZHANG Guangyao, et al. Wind Turbine Fault Diagnosis Based on Transfer Learning and Convolutional Autoencoder with Small-scale Data[J]. Renewable Energy, 2021, 171: 103-115.
[12] 古莹奎, 吴宽, 李成. 基于格拉姆角场和迁移深度残差神经网络的滚动轴承故障诊断[J]. 振动与冲击, 2022, 41(21): 228-237.
GU Yingkui, WU Kuan, LI Cheng. Rolling Bearing Fault Diagnosis Based on Gram Angle Field and Transfer Deep Residual Neural Network[J]. Journal of Vibration and Shock, 2022, 41(21): 228-237.
[13] YANG Jinsong, LIU Jie, XIE Jingsong, et al. Conditional GAN and 2-D CNN for Bearing Fault Diagnosis with Small Samples[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 3525712.
[14] ZHONG Zhidan, LIU Hao, MAO Wentao, et al. Imbalanced Bearing Fault Diagnosis Based on RFH-GAN and PSA-DRSN[J]. IEEE Access, 2023, 11: 131926-131938.
[15] 徐林, 郑晓彤, 付博, 等. 基于改进GAN算法的电机轴承故障诊断方法[J]. 东北大学学报(自然科学版), 2019, 40(12): 1679-1684.
XU Lin, ZHENG Xiaotong, FU Bo, et al. Fault Diagnosis Method of Motor Bearing Based on Improved GAN Algorithm[J]. Journal of Northeastern University(Natural Science), 2019, 40(12): 1679-1684.
[16] 张钊光, 蒋庆磊, 詹瑜滨,等. 基于VAE-GAN数据增强算法的小样本滚动轴承故障分类方法[J]. 原子能科学技术, 2023, 57(S1): 228-237.
ZHANG Zhaoguang, JIANG Qinglei, ZHAN Yubin, et al. Small-sample Rolling Bearing Fault Classification Method Based on VAE-GAN Data Augmentation Algorithm [J]. Atomic Energy Science and Technology, 2023, 57(S1): 228-237.
[17] 郭伟, 邢晓松. 基于改进卷积生成对抗网络的少样本轴承智能诊断方法[J]. 中国机械工程, 2022, 33(19): 2347-2355.
GUO Wei, XING Xiaosong. Intelligent Fault Diagnosis of Bearings with few Samples Based on an Improved Convolutional Generative Adversarial Network[J]. China Mechanical Engineering, 2022, 33(19): 2347-2355.
[18] HU Jie, SHEN Li, SUN Gang. Squeeze-and-excitation Networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 18-23, 2018, Salt Lake City, UT, USA.IEEE,2018: 7132-7141.
[19] YU F, KOLTUN V. Multi-Scale Context Aggregation by Dilated Convolutions [C]//ICLR. 2016.

备注/Memo

备注/Memo:
收稿日期:2025-01-13
作者简介:刘崇睿(1999-),男,硕士研究生。
基金项目:四川省重点研发项目(2023YFG0197); 中央高校基本科研业务费-专题研究项目(2682022ZTPY007); 中国铁道科学研究院集团有限公司基金课题(2023YJ371); 铁路智能勘察关键技术研究与应用(KSNQ232009)
更新日期/Last Update: 2026-01-30