基于SDP和W-DCGAN的滚动轴承故障数据增强方法

摘要/Abstract

摘要： 利用深度生成对抗网络进行滚动轴承小样本故障诊断时,存在生成样本质量不高、模型训练不稳定的缺陷.因此,提出了对称极坐标法(SDP)与Wasserstein距离深度卷积生成对抗网络(W-DCGAN)相结合的方法增强小样本轴承数据. 首先,利用参数优选的SDP转换一维时域信号,得到可输入W-DCGAN的二维图像;然后,采用控制变量法优化生成器和判别器的超参数,确定W-DCGAN基本结构;同时,为了增强W-DCGAN训练过程的稳定性,在判别器中加入谱归一化处理,提升模型生成样本的质量;最后,利用滚动轴承公开数据集对该方法的性能进行了实验验证.结果表明,改进后数据增强方法所生成样本的质量得到了明显提升,生成样本与原始样本的结构相似度和峰值信噪比分别提高了10.12%和12.46%.利用不同方法对滚动轴承故障进行识别,改进后数据增强方法的故障识别准确率最高,达到了99.47%,证明该方法能够有效实现滚动轴承故障小样本数据增强和模式识别.

关键词: 滚动轴承, 小样本, 数据增强, Wasserstein距离深度卷积生成对抗网络, 对称极坐标法

Abstract: When the traditional generative adversarial network is used to solve the fault diagnosis of rolling bearing with small samples, low quality of generated samples and unstable training process are the main defects of the network. Therefore, a new method combining symmetrized dot pattern (SDP) and Wasserstein-distance deep convolutional generative adversarial network (W-DCGAN) is proposed to augment the bearing data with small samples. First, the SDP with optimized parameters is used to transform the one-dimensional time-domain signal to the two-dimensional image suitable for inputting into W-DCGAN. Then, the controlled variable method is used to optimize the hyperparameters of the generator and discriminator to determine the basic structure of the W-DCGAN. Meantime, in order to enhance the stability of training process of W-DCGAN, spectral normalization is added into discriminator to improve the quality of generated samples. The performance of the method is verified by an open rolling bearing dataset. The results show that the quality of the generated samples of the proposed method is significantly improved. The structural similarity and peak signal-to-noise ratio of the generated samples relative to the original samples increase by 10.12% and 12.46%, respectively. For the rolling bearing fault identification, the proposed method has the highest accuracy of fault identification, reaching 99.47%, indicating its effectiveness in small sample data augmentation andpattern recognition for rolling bearing fault.

Key words: rolling bearing, small sample, data augmentation, W-DCGAN, symmetrized dot pattern

中图分类号:

宋慧贤, 邓林峰, 郑玉巧. 基于SDP和W-DCGAN的滚动轴承故障数据增强方法[J]. 兰州理工大学学报, 2025, 51(6): 55-64.

SONG Hui-xian, DENG Lin-feng, ZHENG Yu-qiao. Data augmentation of rolling bearing faults based on SDP and W-DCGAN[J]. Journal of Lanzhou University of Technology, 2025, 51(6): 55-64.

参考文献

[1] WANG Z L,YANG J H,GUO Y.Unknown fault feature extraction of rolling bearings under variable speed conditions based on statistical complexity measures[J].Mechanical Systems and Signal Processing,2022,172: 108964.
[2] 何强,唐向红,李传江,等.负载不平衡下小样本数据的轴承故障诊断[J].中国机械工程,2021,32(10):1164-1171.
[3] ZHANG K Y,CHEN J L,ZHANG T C,et al.A compact convolutional neural network augmented with multiscale feature extraction of acquired monitoring data for mechanical intelligent fault diagnosis[J].Journal of Manufacturing Systems,2020,55:273-284.
[4] ZHANG T C,CHEN J L,LI F D,et al.Intelligent fault diagnosis of machines with small and imbalanced data:a state-of-the-art review and possible extensions[J].ISA Transactions,2022,119:152-171.
[5] 朱克凡,王杰贵,刘有军.小样本条件下基于数据增强和WACGAN的雷达目标识别算法[J].电子学报,2020,48(6):1124-1131.
[6] CHEN R X,ZHU J K,HU X L,et al.Fault diagnosis method of rolling bearing based on multiple classifier ensemble of the weighted and balanced distribution adaptation under limited sample imbalance[J].ISA Transactions,2021,114:434-443.
[7] 王琦,邓林峰,赵荣珍.基于改进一维卷积神经网络的滚动轴承故障识别[J].振动与冲击,2022,41(3):216-223.
[8] ZHANG H,XU T,LI H S,et al.StackGAN++:realistic image synthesis with stacked generative adversarial networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2019,41(8):1947-1962.
[9] CHEN Y,ZHAO Y,JIA W,et al.Adversarial-learning based image-to-image transformation:a survey[J].Neurocomputing,2020,411:468-486.
[10] 罗佳,黄晋英.生成式对抗网络研究综述[J].仪器仪表学报,2019,40(3):74-84.
[11] MINAEE S,BOYKOV Y Y,PORIKLI F,et al.Image segmentation using deep learning:a survey[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2022,44(7):3523-3542.
[12] 肖雄,肖宇雄,张勇军,等.基于二维灰度图的数据增强方法在电机轴承故障诊断的应用研究[J].中国电机工程学报,2021,52(2):738-748.
[13] 李庆旭,王巧华,马美湖.基于生成对抗网络的禽蛋图像数据生成研究[J].农业机械学报,2021,52(2):236-245.
[14] 崔昊杨,韩奕,张驯,等.基于改进U-Net结构生成对抗网络的运动模糊绝缘子图像复原方法[J].电网技术,2023,47(6):2594-2604.
[15] ZHU X X,HOU D N,ZHOU P,et al.Rotor fault diagnosis using a convolutional neural network with symmetrized dot pattern images[J].Measurement,2019,138:526-535.
[16] TANG Y,ZHANG X F,QIN G J,et al.Graph cardinality preserved attention network for fault diagnosis of induction motor under varying speed and load condition[J].IEEE Transactions on Industrial Informatics,2022,18(6):3702-3712.
[17] WANG H,XU J W,YAN R Q,et al.A new intelligent bearing fault diagnosis method using SDP representation and SE-CNN[J].IEEE Transactions on Instrumentation and Measurement,2020,69(5):2377-2389.
[18] SUN Y J,LI S H.Bearing fault diagnosis based on optimal convolution neural network[J].Measurement,2022,190:110702.
[19] LUO J,HUANG J Y,MA J C,et al.An evaluation method of conditional deep convolutional generative adversarial networks for mechanical fault diagnosis[J].Journal of Vibration and Control,2022,28(11/12):1379-1389.
[20] 李东东,刘宇航,赵阳,等.基于改进生成对抗网络的风机行星齿轮箱故障诊断方法[J].中国电机工程学报,2021,41(21):7496-7507.
[21] 卢锦玲,张祥国,张伟,等.基于改进辅助分类生成对抗网络的风机主轴承故障诊断[J].电力系统自动化,2021,45(7):148-154.
[22] ZHANG C Z,WANG W,LI H.Tool wear prediction method based on symmetrized dot pattern and multi-covariance gaussian process regression[J].Measurement,2022,189:110466.
[23] MIYATO T,KATAOKA T,KOYAMA M,et al.Spectral normalization for generative adversarial networks[C] //In Proceeding of 6th International Conference on Learning Representations.[S.l.]:arXiv,2018:1-26.
[24] NEUPANE D,SEOK J.Bearing fault detection and diagnosis using case western reserve university dataset with deep learning approaches:a review[J].IEEE Access,2020,8:93155-93178.
[25] 史建超,胡正伟,贺冬梅,等.基于深度学习的电力线信道传输特性识别方法[J].电网技术,2019,43(12):4283-4290.
[26] XU M Q,WANG Y Q.An imbalanced fault diagnosis method for rolling bearing based on semi-supervised conditional generative adversarial network with spectral normalization[J].IEEE Access,2021,9:27736-27747.
[27] 曹洁,张玉林,王进花,等.基于VMD和SVPSO-BP的滚动轴承故障诊断[J].太阳能学报,2022,43(9):294-301.
[28] 郑直,张华钦,潘月.基于改进鲸鱼算法优化LSTM的滚动轴承故障诊断[J].振动与冲击,2021,40(7):274-280.