基于SCACGAN的小样本齿轮箱故障诊断

王进花; 刘秦玮; 曹洁; 陈莉

doi:10.13700/j.bh.1001-5965.2023.0819

基于SCACGAN的小样本齿轮箱故障诊断

doi: 10.13700/j.bh.1001-5965.2023.0819

王进花^1, ,,
刘秦玮¹,
曹洁^{1, 2, 3},
陈莉²

1.
兰州理工大学电气工程与信息工程学院，兰州 730050
2.
兰州城市学院信息工程学院，兰州 730070
3.
甘肃省制造信息工程研究中心，兰州 730050

基金项目:

国家自然科学基金(62063020,61763028)；国家重点研发计划(2020YFB1713600)；甘肃省自然科学基金(20JR5RA463)

详细信息

通讯作者:
E-mail：wjh0615@lut.edu.cn

中图分类号: TP277；TH133.33
计量
- 文章访问数: 3
- HTML全文浏览量: 0
- PDF下载量: 9
- 被引次数: 0
出版历程
- 收稿日期: 2023-12-18
- 录用日期: 2024-03-08
- 网络出版日期: 2024-04-30
- 整期出版日期: 2026-03-31

Fault diagnosis of gearbox with small-sample based on SCACGAN

WANG Jinhua^{1
, ,},
LIU Qinwei¹,
CAO Jie^{1, 2, 3},
CHEN Li²

1.
School of Electrical Engineering and Information Engineering，Lanzhou University of Technology，Lanzhou 730050，China
2.
School of Information Engineering，Lanzhou City University，Lanzhou 730070，China
3.
Gansu Manufacturing Information Engineering Research Center，Lanzhou 730050，China

Funds:

National Natural Science Foundation of China (62063020,61763028); National Key Research and Development Program of China (2020YFB1713600); Natural Science Foundation of Gansu Province (20JR5RA463)

More Information

Corresponding author: E-mail：wjh0615@lut.edu.cn

摘要

摘要:
针对辅助分类器生成对抗网络(ACGAN)在小样本齿轮箱故障诊断过程中，生成故障样本缺乏多样性，且质量较差，导致诊断准确度不高的问题，提出一种基于自校正辅助分类器生成对抗网络(SCACGAN)的齿轮箱故障诊断方法。在辅助分类器生成对抗网络中引入一个独立的分类器，改善判别器输出错误对生成样本质量造成的不良影响，并对不同齿轮箱样本的健康状况进行分类；采用最小二乘函数，提高模型的生成能力和分类能力，改善训练过程中生成样本质量不高的问题；在生成器中引入自校正卷积神经网络，增强故障特征获取的能力。实验结果表明：在小样本条件下，所提方法能够生成质量较好的故障样本，从而提高了齿轮箱的故障诊断准确度。
- 齿轮箱 /
- 小样本 /
- 辅助分类器生成对抗网络 /
- 自校正卷积神经网络 /
- 故障诊断
Abstract:
A new method for gearbox fault diagnosis based on the self-correcting auxiliary classifier generative adversarial networks (SCACGAN) is suggested in response to the limited diversity and low quality of fault samples produced by the auxiliary classifier generative adversarial networks (ACGAN) during the small-sample gearbox fault diagnosis process, which subsequently results in low diagnostic accuracy. Firstly, an independent classifier is introduced into the auxiliary classifier generative adversarial network to mitigate the adverse impact of discriminator output errors on the quality of generated samples, and to classify the health status of different gearbox samples. Secondly, the problem of low-quality generated samples during the training phase is addressed by using the least squares function to improve the model’s generation and classification skills. Lastly, a self-correcting convolutional neural network is integrated into the generator to enhance the capability of fault feature acquisition. Experimental results demonstrate that under small-sample conditions, the proposed approach is capable of generating higher-quality fault samples, thereby improving the accuracy of gearbox fault diagnosis.
- gearbox /
- small-sample /
- auxiliary classifier generative adversarial networks /
- self-correcting convolutional neural networks /
- fault diagnosis

HTML全文

图 1 ACGAN结构

Figure 1. Architecture of ACGAN

下载: 全尺寸图片幻灯片

图 2 SCConv结构

Figure 2. Architecture of SCConv

下载: 全尺寸图片幻灯片

图 3 SCACGAN生成器结构

Figure 3. Architecture of SCACGAN generator

下载: 全尺寸图片幻灯片

图 4 SCACGAN的判别器和分类器

Figure 4. Discriminator and classifier of SCACGAN

下载: 全尺寸图片幻灯片

图 5 SCACGAN结构

Figure 5. Architectural of SCACGAN

下载: 全尺寸图片幻灯片

图 6 SCACGAN的小样本齿轮箱故障诊断流程图

Figure 6. Process diagram of SCACGAN for small-sample gearbox fault diagnosis

下载: 全尺寸图片幻灯片

图 7 康涅狄格大学的齿轮箱结构

Figure 7. Gearbox structure at the University of Connecticut

下载: 全尺寸图片幻灯片

图 8 齿轮的9个不同健康状况

Figure 8. Nine different health conditions of gears

下载: 全尺寸图片幻灯片

图 9 不同健康状况下的灰度图

Figure 9. Grayscale images under different health conditions

下载: 全尺寸图片幻灯片

图 10 基于SSIM的对比结果

Figure 10. Contrast results based on SSIM

下载: 全尺寸图片幻灯片

图 11 混淆矩阵图

Figure 11. Confusion matrix plot

下载: 全尺寸图片幻灯片

图 12 不同模型的特征散点图

Figure 12. Scatter plot of features for different models

下载: 全尺寸图片幻灯片

图 13 不同样本数量下的平均准确度和准确度标准差

Figure 13. Average accuracy and standard deviation under different sample sizes

下载: 全尺寸图片幻灯片

表 1 SCACGAN的生成器结构参数

Table 1. Generator structural parameters of SCACGAN

网络层	结构参数	H×W×N
Dense	2 048	2×2×512
Conv1	(128,5,2)	4×4×128
Conv2	(64,5,2)	8×8×64
SCConv	(32,5)（经过K₁）	8×8×64
	(32,5) （经过K₂）
	(32,5) （经过K₃）
	(32,5)（经过K₄）
Conv3	(8,5,2)	16×16×8
Conv4	(1,3,2)	32×32×1

下载: 导出CSV

表 2 SCACGAN的判别器和分类器结构参数

Table 2. Discriminator and classifier structural parameters of SCACGAN

模型	网络层	结构参数	输出尺寸
判别器	Conv1	(16,5,2)	16×16×16
	Conv2	(32,3,2)	8×8×32
	Conv3	(64,3,2)	4×4×64
	Conv4	(128,3,2)	2×2×128
	Conv5	(256,3,1)	2×2×256
	Dense	1	1
分类器	Conv1	(16,3,2)	16×16×16
	Conv2	(32,3,2)	8×8×32
	Conv3	(64,3,2)	4×4×64
	Conv4	(128,3,2)	2×2×128
	Conv5	(256,3,1)	2×2×256
	Dense	256	256
	Dense	128	128
	Dense	9	9

下载: 导出CSV

表 3 齿轮样品详情

Table 3. Gear sample details

齿轮的健康状态	分类标签
健康	0
缺齿	1
齿根裂纹	2
剥落	3
碎屑尖端1	4
碎屑尖端2	5
碎屑尖端3	6
碎屑尖端4	7
碎屑尖端5	8

下载: 导出CSV

表 4 对比模型参数

Table 4. Comparison model parameters

模型	模块名称	学习率
LSTM^[23]		0.01
CNN^[24]		0.01
CGAN^[25]	生成器	0.0002
CGAN^[25]	判别器	0.0001
ACGAN^[18]	生成器	0.0003
ACGAN^[18]	判别器	0.0002
WAC-GAN^[19]	生成器	0.0003
WAC-GAN^[19]	判别器	0.0002
SCACGAN	生成器	0.0001
	判别器	0.0001
	分类器	0.00001

下载: 导出CSV

参考文献(25)

[1]	ZHANG M, LI D C, WANG K S, et al. An adaptive order-band energy ratio method for the fault diagnosis of planetary gearboxes[J]. Mechanical Systems and Signal Processing, 2022, 165: 108336.
[2]	XUE S, HOWARD I, WANG C S, et al. Dynamic modelling of the gear system under non-stationary conditions using the iterative convergence of the tooth mesh stiffness[J]. Engineering Failure Analysis, 2022, 131: 105908.
[3]	RUAN D W, WANG J, YAN J P, et al. CNN parameter design based on fault signal analysis and its application in bearing fault diagnosis[J]. Advanced Engineering Informatics, 2023, 55: 101877.
[4]	JIA L S, CHOW T W S, YUAN Y X. GTFE-Net: a gramian time frequency enhancement CNN for bearing fault diagnosis[J]. Engineering Applications of Artificial Intelligence, 2023, 119: 105794.
[5]	WANG S D, LIU Z B, JIA Z, et al. Composite fault diagnosis of analog circuit system using chaotic game optimization-assisted deep ELM-AE[J]. Measurement, 2022, 202: 111826.
[6]	YANG D G, KARIMI H R, SUN K K. Residual wide-kernel deep convolutional auto-encoder for intelligent rotating machinery fault diagnosis with limited samples[J]. Neural Networks, 2021, 141: 133-144.
[7]	LEI Y X, KARIMI H R, CHEN X F. A novel self-supervised deep LSTM network for industrial temperature prediction in aluminum processes application[J]. Neurocomputing, 2022, 502: 177-185.
[8]	ZHANG S Y, QIU T. Semi-supervised LSTM ladder autoencoder for chemical process fault diagnosis and localization[J]. Chemical Engineering Science, 2022, 251: 117467.
[9]	JIN Z Z, HE D Q, WEI Z X. Intelligent fault diagnosis of train axle box bearing based on parameter optimization VMD and improved DBN[J]. Engineering Applications of Artificial Intelligence, 2022, 110: 104713.
[10]	GAO S Z, XU L T, ZHANG Y M, et al. Rolling bearing fault diagnosis based on SSA optimized self-adaptive DBN[J]. ISA Transactions, 2022, 128: 485-502.
[11]	LI Y J, WANG Y B, ZHANG Y, et al. Diagnosis of inter-turn short circuit of permanent magnet synchronous motor based on deep learning and small fault samples[J]. Neurocomputing, 2021, 442: 348-358.
[12]	MAO W T, FENG W S, LIU Y M, et al. A new deep auto-encoder method with fusing discriminant information for bearing fault diagnosis[J]. Mechanical Systems and Signal Processing, 2021, 150: 107233.
[13]	WANG Z S, XUAN J P. Intelligent fault recognition framework by using deep reinforcement learning with one dimension convolution and improved actor-critic algorithm[J]. Advanced Engineering Informatics, 2021, 49: 101315.
[14]	HE Z Y, SHAO H D, DING Z Y, et al. Modified deep autoencoder driven by multisource parameters for fault transfer prognosis of aeroengine[J]. IEEE Transactions on Industrial Electronics, 2021, 69(1): 845-855.
[15]	ZHOU F N, YANG S, FUJITA H, et al. Deep learning fault diagnosis method based on global optimization GAN for unbalanced data[J]. Knowledge-Based Systems, 2020, 187: 104837.
[16]	DU Z M, CHEN K, CHEN S L, et al. Deep learning GAN-based data generation and fault diagnosis in the data center HVAC system[J]. Energy and Buildings, 2023, 289: 113072.
[17]	DU W Y, YANG J P, MENG G L. Fault diagnosis for dynamic system based on the independent latent space reconstruction of generative adversarial network[J]. Journal of Process Control, 2023, 125: 28-40.
[18]	ODENA A, OLAH C, SHLENS J. Conditional image synthesis with auxiliary classifier gans[C]//Proceedings of the International Conference on Machine Learning. Sydney: PMLR, 2017: 2642-2651.
[19]	SUN Q, PENG F, YU X H, et al. Data augmentation strategy for power inverter fault diagnosis based on Wasserstein distance and auxiliary classification generative adversarial network[J]. Reliability Engineering & System Safety, 2023, 237: 109360.
[20]	许小伟, 韦道明, 严运兵, 等. 基于改进ACGAN的永磁同步电机数据扩张方法[J]. 哈尔滨工业大学学报, 2023, 55(10): 114-121. XU X W, WEI D M, YAN Y B, et al. Data expansion method of permanent magnet synchronous motor based on improved ACGAN[J]. Journal of Harbin Institute of Technology, 2023, 55(10): 114-121(in Chinese).
[21]	李慧芳, 徐光浩, 黄双喜. 基于主动生成式过采样和深度堆叠网络的轴承故障诊断[J]. 计算机集成制造系统, 2023, 29(1): 146-159. LI H F, XU G H, HUANG S X. Active generative oversampling and deep stacking network based bearing fault diagnosis approach[J]. Computer Integrated Manufacturing Systems, 2023, 29(1): 146-159 (in Chinese).
[22]	MAO X D, LI Q, XIE H R, et al. Least squares generative adversarial networks[C]//Proceedings of the IEEE International Conference on Computer Vision. Piscataway: IEEE Press, 2017: 2813-2821.
[23]	LEI J H, LIU C, JIANG D X. Fault diagnosis of wind turbine based on long short-term memory networks[J]. Renewable Energy, 2019, 133: 422-432.
[24]	LI Y, ZHANG Y, LIU W, et al. A fault pattern and convolutional neural network based single-phase earth fault identification method for distribution network [C]//Proceedings of the IEEE Innovative Smart Grid Technologies-Asia. Piscataway: IEEE press, 2019: 838-843.
[25]	MIRZA M, OSINDERO S. Conditional generative adversarial nets[EB/OL]. (2014-12-06)[2023-12-16]. https://arxiv.org/abs/1411.1784.