Imbalanced data enhancement method based on improved DCGAN and its application

Abstract

Machinery operates well under normal conditions in most cases; far fewer samples are collected in a fault state (minority samples) than in a normal state, resulting in an imbalance of samples. Common machine learning algorithms such as deep neural networks require a significant amount of data during training to avoid overfitting. These models often fail to detect minority samples when the input samples are imbalanced, which results in missed diagnoses of equipment faults. As an effective method to enhance minority samples, Deep Convolution Generative Adversarial Network (DCGAN) does not fundamentally address the problem of unstable Generative Adversarial Network (GAN) training. This study proposes an improved DCGAN model with improved stability and sample balance for achieving greater classification accuracy over minority samples. First, spectral normalization is performed on each convolutional layer, improving stability in the DCGAN discriminator. Then, the improved DCGAN model is trained to generate new samples that are different from the original samples but with a similar distribution when the Nash equilibrium is reached. Four indices—Inception Score (IS), Fréchet Inception Distance Score (FID), Peak Signal to Noise Ratio (PSNR), and Structural Similarity (SSIM)—were used to quantitatively evaluate of the generated images. Finally, the Balance Degree of Samples (BDS) index was proposed, and the new samples are proportionally added to the original samples to improve sample balance, resulting in the formation of several groups of datasets with different balance degrees, and Convolutional Neural Network (CNN) models are used to classify these samples. With experimental analysis on the reciprocating compressor, the variance of lost data is found to be less than 1% of the original value, representing an increase in stabilityof the model to generate diverse and high-quality sample images, as compared with that of the unmodified model. The classification accuracy exceeds 95% and tends to remain stable when the balance degree of samples is greater than 80%.

Keywords

Imbalanced data enhancement fault diagnosis DCGAN CNN

Get full access to this article

View all access options for this article.

References

Wang

Z.X.

, FLOWGAN: Research on key technology of encrypted traffic identification based on generative adversarial network, Master Dissertation, Nanjing University Of Posts And Telecommunications, (2020).

Chawlan

, Bowyer

K.W.

and Hall

O.L.

, SMOTE: synthetic minority over-sampling technique, Journal of Artificial Intelligence Research (2002), 321–357.

Xie

M.Y.

, Methods for Reciprocating Compressor Multi-Fault Identification under Imbalanced Datasets, Master Dissertation, China University of Petroleum (Beijing), (2018).

Xiang

H.Y.

and Yang

, Survey on Imbalanced Data Mining Methods, Computer Engineering and Applications (2019), 1–16.

Zhang

X.F.

and Gang

, Data augmentation method based on generative adversarial network, Computer Systems and Applications (2019), 201–206.

Lan

, Shen

H.F.

and Wang

, Dataset enhancement method based on improved deep convolutional generative adversarial networks, Journal of Computer Applications (2021), 1–11.

Mao

X.D.

, Li

and Xie

H.R.

, Least squares generative adversarial networks, Proceedings of IEEE International Conference on Computer Vision (ICCV), (2017).

Goodfellow

I.J.

, Pouget-Abadie

and Mirza

, Generative adversarial nets, Conference and Workshop on Neural Information Processing Systems (2014), 2672–2680.

Luc

, Couprie

and Chintala

, Semantic segmentation using adversarial networks, Proceedings of the 30th Conference on Neural Information Processing Systems (2016).

10.

Zhou

, Research on Image Generation Method Based on Deep Learning, Master Dissertation, Lanzhou University, (2019).

11.

Radford

, Metz

and Chintala

, Unsupervised representation learning with deep convolutional generative adversarial networks, International Conference on Learning Representations (2016), 1–16.

12.

Arjovsky

, Chintala

and Bottou

, Wasserstein generative adversarial networks, Proceedings of the 34th International Conference on Machine Learning Princeton, (2017), 298321.

13.

Gulrajani

, Ahmed

and Arjovsky

, Improved training of Wasserstein GANs[J], ar Xiv:1704.00028, (2017).

14.

Liang

J.J.

, Wei

J.J.

and Jiang

Z.F.

, Generative Adversarial Networks GAN Overview, Journal of Frontiers of Computer Science and Technology (2020), 1–17.

15.

Zhang

Z.H.

, Zeng

Y.B.

and Lu

, Spectral bounding: Strictlysatisfying the 1-Lipschitz property for generative adversarialnetworks[J], Pattern Recognition 105 2020, 1–10.

16.

Tan

H.W.

, Zhou

L.Y.

and Wang

G.D.

, Instability analysis for generative adversarial networks and its solving techniques, Scientia Sinica(Informationis) (2021), 1–16.

17.

Miyato

, Kataoka

and Koyama

, Spectral Normalization for Generative Adversarial Networks[EB/OL], https://arxiv.org/pdf/1802.05957.pdf, (2020).

18.

N.N.

and Zheng

, Multi-loss ensemble and spectral normalization for image super-resolution, Application Research of Computers (2020), 2531–2535.

19.

Z.P.

, Guo

L.L.

and Zhu

M.C.

, Improved image registration using feature points combined with image entropy, Infrared and Laser Engineering (2013), 2846–2852.

20.

Scaman

and Virmaux

, Lipschitz regularity of deep neural networks: Analysis and efficient estimation, Proceedings of the 32nd Conference on Neural Information Processing Systems (2018), 3835–3844.

21.

Zhang

Z.X.

, Yuan

J.Q.

and Guo

H.L.

, Multi-discriminator Co-operation Framework: Spectral Normalized Generative Adversarial Networks for High Quality Generated Images, Journal of Chinese Computer Systems (2021), 201–207.

22.

Gan

, Shen

H.F.

and Wang

, Simulation Method of Target SAR Image Based on Spectral Normalization Generative Adversarial Network, Computer and Modernization (2021), 1–11.

23.

, Research on heterogeneous features fusion and online imbalance classification for bearing fault diagnosis, Master Dissertation, Henan Normal University, (2017).

24.

Zhong

S.S.

, Li

and Zhang

Y.J.

, Fault diagnosis of civil aero-engine driven by unbalanced amples based on DBN, Journal of Aerospace Power (2019), 708–716.

25.

Borji

, Pros and cons of GAN evaluation measures, Computer Vision and Image Understanding (2019), 41–65.

26.

Salimans

, Goodfellow

and Zaremba

, Improved techniques for training GANs, Advances in Neural Information Processing Systems, Curran Associates (2016), 2234–2242.

27.

Tran

N.T.

, Tran

V.H.

and Nguyen

N.B.

, On Data Augmentation for GAN Training, IEEE Transactions on Image Processing (2021), 1882–1897.

28.

Zhang

, Xu

G.H.

and Han

Z.Z.

, Data Augmentation for Motor Imagery Signal Classification Based on a Hybrid Neural Network, Sensors (2020), 1–20.

29.

Huynh-Thu

and Ghanbari

, Scope of validity of PSNR in image/video quality assessment, Electronics Letters (2008), 800–801.

30.

Wang

, Bovik

A.C.

and Sheikh

H.R.

, Image quality assessment: from error visibility to structural similarity, IEEE Transactions on Image Processing (2004), 600–612.

31.

Zhong

G.Q.

, Gao

and Liu

Y.B.

, Generative adversarial networks with decoder–encoder output noises, Neural Networks (2020), 19–28.

32.

Szegedy

, Vanhoucke

and Ioffe

, Rethinking the Inception Architecture for Computer Vision, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016), 2818–2826.

33.

Russakovsky

, Deng

and Su

, ImageNet Large Scale Visual Recognition Challenge, International Journal of Computer Vision (2015), 211–252.

34.

Szegedy

, Liu

and Jia

, Going Deeper with Convolutions, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). USA: IEEE Computer Society, (2015), 1–9.

35.

, Zhang

and Ren

, Deep Residual Learning for Image Recognition, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), USA: IEEE Computer Society, (2016), 770–778.