GAN-based data augmentation of time series for fault diagnosis in railway track

Abstract

Supervised learning classification has arisen as a powerful tool to perform data-driven fault diagnosis in dynamical systems, achieving astonishing results. This approach assumes the availability of extensive, diverse and labeled data corpora for training. However, in some applications it may be difficult or not feasible to obtain a large and balanced dataset including enough representative instances of the fault behaviors of interest. This fact leads to the issues of data scarcity and class imbalance, greatly affecting the performance of supervised learning classifiers. Datasets from railway systems are usually both, scarce and imbalanced, turning supervised learning-based fault diagnosis into a highly challenging task. This article addresses time-series data augmentation for fault diagnosis purposes and presents two application cases in the context of railway track. The case studies employ generative adversarial networks (GAN) schemes to produce realistic synthetic samples of geometrical and structural track defects. The goal is to generate samples that enhance fault diagnosis performance; therefore, major attention was paid not only in the generation process, but also in the synthesis quality assessment, to guarantee the suitability of the samples for training of supervised learning classification models. In the first application, a convolutional classifier achieved a test accuracy of 87.5% for the train on synthetic, test on real (TSTR) scenario, while, in the second application, a fully-connected classifier achieved 96.18% in test accuracy for TSTR. The results indicate that the proposed augmentation approach produces samples having equivalent statistical characteristics and leading to a similar classification behavior as real data.

Keywords

Data augmentation / Time series / Generative adversarial networks / Fault diagnosis / Predictive maintenance / Railway systems

References

[1] Fernández-Bobadilla HA, Martin U. Modern tendencies in vehicle-based condition monitoring of the railway track. IEEE Trans Instrum Meas, 2023, 72: 1-44

[2] Goodfellow I, Pouget-Abadie J, Mirza M et al (2014) Generative adversarial nets. Adv Neural Inf Process Syst 27

[3] Iglesias G, Talavera E, González-Prieto Á, et al.. Data augmentation techniques in time series domain: a survey and taxonomy. Neural Comput Appl, 2023, 35(14): 10123-10145

[4] Iwana BK, Uchida S. An empirical survey of data augmentation for time series classification with neural networks. PLoS ONE, 2021, 16(7): 0254841

[5] Camacho Alcocer D, Le T, Rapp S, et al.. Light rail ballasted track geometry quality evaluation using track recording car data. Comput Railw XV: Railw Eng Des Operat, 2016, 162: 303

[6] Cabral R, Ribeiro D, Rakoczy A. Ribeiro D, Montenegro PA, Andersson A, Martínez-Rodrigo MD. Engineering the future: a deep dive into remote inspection and reality capture for railway infrastructure digitalization. Digital railway infrastructure, 2024, Cham, Springer229-256

[7] Hajizadeh S, Núñez A, Tax DM. Semi-supervised rail defect detection from imbalanced image data. IFAC-PapersOnLine, 2016, 49(378-83

[8] Yao D, Sun Q, Yang J, et al.. Railway fastener fault diagnosis based on generative adversarial network and residual network model. Shock Vib, 2020, 2020(1): 8823050

[9] Chen Y, Tong S, Lu X et al (2021) A semi-supervised railway foreign object detection method based on GAN. In: Proceedings of the 5th international conference on computer science and application engineering, pp 1–5

[10] Yang H, Li X, Guo Y, et al.. Rt-GAN: GAN based architecture for precise segmentation of railway tracks. Appl Sci, 2022, 12(2312044

[11] Liu R, Liu W, Zheng Z, et al.. Anomaly-GAN: a data augmentation method for train surface anomaly detection. Expert Syst Appl, 2023, 228: 120284

[12] Kulevome DKB, Wang H, Cobbinah BM, et al.. Effective time-series data augmentation with analytic wavelets for bearing fault diagnosis. Expert Syst Appl, 2024, 249: 123536

[13] Sobie C, Freitas C, Nicolai M. Simulation-driven machine learning: bearing fault classification. Mech Syst Sign Process, 2018, 99: 403-419

[14] Gao Y, Liu X, Xiang J. FEM simulation-based generative adversarial networks to detect bearing faults. IEEE Trans Industr Inf, 2020, 16(74961-4971

[15] Luo J, Huang J, Li H. A case study of conditional deep convolutional generative adversarial networks in machine fault diagnosis. J Intell Manuf, 2021, 32(2): 407-425

[16] Liu S, Jiang H, Wu Z, et al.. Data synthesis using deep feature enhanced generative adversarial networks for rolling bearing imbalanced fault diagnosis. Mech Syst Sign Process, 2022, 163: 108139

[17] Smith KE, Smith AO (2020) Conditional GAN for time series generation. arXiv preprint arXiv:2006.16477

[18] Shi D, Ye Y, Gillwald M, et al.. Robustness enhancement of machine fault diagnostic models for railway applications through data augmentation. Mech Syst Sign Process, 2022, 164: 108217

[19] Li M, Hei X, Ji W, et al.. A fault-diagnosis method for railway turnout systems based on improved autoencoder and data augmentation. Sensors, 2022, 22(23): 9438

[20] Shangguan A, Xie G, Fei R, et al.. Train wheel degradation generation and prediction based on the time series generation adversarial network. Reliab Eng Syst Saf, 2023, 229: 108816

[21] Wang Y, Cai X, Tang X, et al.. HSRA-Net: intelligent detection network of anomaly monitoring data in high-speed railway. IEEE Trans Intell Transp Syst, 2024

[22] Yoon J, Jarrett D, Schaar M (2019) Time-series generative adversarial networks. Adv. Neural Inf Process Syst 32

[23] Schenkendorf R, Dutschk B (2016) Improved railway track irregularities classification by a model inversion approach. In: Third European conference of the prognostics and health management society 2016(July), pp 358–365

[24] Baasch B, Roth M, Groos JC. In-service condition monitoring of rail tracks. Internationales Verkehrswesen, 2018, 70: 76-79

[25] Abanda A, Mori U, Lozano JA. A review on distance based time series classification. Data Min Knowl Disc, 2019, 33(2378-412

[26] Gatta F, Giampaolo F, Prezioso E, et al.. Neural networks generative models for time series. J King Saud Univ-Comput Inf Sci, 2022, 34(10): 7920-7939

[27] Shumway RH, Stoffer DS. Time series analysis and its applications: with R examples, 2017, New York, Springer

[28] Park DS, Chan W, Zhang Y, et al.. SpecAugment: a simple data augmentation method for automatic speech recognition. Proc Interspeech, 2019, 2019: 2613-2617

[29] Mirza M, Osindero S (2014) Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784

[30] Arjovsky M, Chintala S, Bottou L (2017) Wasserstein generative adversarial networks. In: International conference on machine learning, PMLR, pp 214–223

[31] Gulrajani I, Ahmed F, Arjovsky M et al (2017) Improved training of Wasserstein GANs. arXiv preprint arXiv:1704.00028

[32] Harshvardhan G, Gourisaria MK, Pandey M, et al.. A comprehensive survey and analysis of generative models in machine learning. Comput Sci Rev, 2020, 38: 100285

[33] Xu B, Zhou D, Li W. Image enhancement algorithm based on GAN neural network. IEEE Access, 2022, 10: 36766-36777

[34] Hazra D, Byun Y-C. Synsiggan: generative adversarial networks for synthetic biomedical signal generation. Biology, 2020, 9(12): 441

[35] Vaswani A, Shazeer N, Parmar N et al (2017) Attention is all you need. Adv Neural Inf Process Syst 30

[36] Srinivasan P, Knottenbelt WJ (2022) Time-series transformer generative adversarial networks. arXiv preprint arXiv:2205.11164

[37] Garg VK, Dukkipati RV. Dynamics of railway vehicle systems, 1984, Ontario, Academic Press Canada

[38] Knothe K, Stichel S. Schienenfahrzeugdynamik [Railway Vehicle Dynamics], 2003, Berlin, Springer

[39] Zhai W. Vehicle-track coupled dynamics: theory and applications, 2020, Singapore, Springer

[40] Podwórna M. Modelling of random vertical irregularities of railway tracks. Int J Appl Mech Eng, 2015, 20(3647-655

[41] Lei X. Track irregularity power spectrum and numerical simulation. High speed railway track dynamics, 2017, Singapore, Springer137160

[42] Haigermoser A, Luber B, Rauh J, et al.. Road and track irregularities: measurement, assessment and simulation. Veh Syst Dyn, 2015, 53(7): 878-957

[43] Fries R, Coffey B (1990) A state-space approach to the synthesis of random vertical and crosslevel rail irregularities. J Dyn Syst Meas Control

[44] Jesussek M, Ellermann K. Fault detection and isolation for a railway vehicle by evaluating estimation residuals. Proc Iutam, 2015, 13: 14-23

[45] Panunzio AM, Puel G, Cottereau R, et al.. Construction of a stochastic model of track geometry irregularities and validation through experimental measurements of dynamic loading. Veh Syst Dyn, 2017, 55(3): 399-426

[46] Perrin G, Soize C, Duhamel D, et al.. Track irregularities stochastic modeling. Probab Eng Mech, 2013, 34: 123-130

[47] Fernández-Bobadilla HA, Martin U (2024) GAN-based data augmentation of railway track irregularities for fault diagnosis. In: 2024 International joint conference on neural networks (IJCNN), pp 1–9

[48] Sharma A, Singh PK, Chandra R. Smotified-GAN for class imbalanced pattern classification problems. IEEE Access, 2022, 10: 30655-30665

[49] Luo T-J, Fan Y, Chen L, et al.. EEG signal reconstruction using a generative adversarial network with Wasserstein distance and temporal-spatial-frequency loss. Front Neuroinform, 2020, 14: 15

[50] Turner R, Hung J, Frank E et al (2019) Metropolis-Hastings generative adversarial networks. In: International conference on machine learning, PMLR, pp 6345–6353

[51] Heusel M, Ramsauer H, Unterthiner T et al (2017) GANs trained by a two time-scale update rule converge to a local Nash equilibrium. Adv. Neural Inf Process Syst 30

[52] Salimans T, Goodfellow I, Zaremba W et al (2016) Improved techniques for training GANs. Adv Neural Inf Process Syst 29

[53] Esteban C, Hyland SL, Rätsch G (2017) Real-valued (medical) time series generation with recurrent conditional GANs. arXiv preprint arXiv:1706.02633

[54] Năstăsescu G-S, Cercel D-C (2022) Conditional Wasserstein GAN for energy load forecasting in large buildings. In: 2022 International joint conference on neural networks (IJCNN), IEEE, pp 1–8

[55] Rapp S, Martin U, Strähle M, et al.. Track-vehicle scale model for evaluating local track defects detection methods. Transp Geotech, 2019, 19: 9-18

[56] Bahamon-Blanco S, Rapp S, Rupp C et al (2019) Recognition of track defects through measured acceleration-part 1. In: IOP conference series: materials science and engineering, IOP Publishing, vol. 615, p 012121

Funding

Deutsche Forschungsgemeinschaft(515687155)

RIGHTS & PERMISSIONS

The Author(s)