A Local-Global Interactive Time-Series Generation Method for Imbalanced Operating Status Assessment
DOI:
https://doi.org/10.54097/1wme0740Keywords:
Deep learning, operating status assessment, class imbalance, data augmentationAbstract
To address the imbalance of operating-status samples in multivariate time series from complex industrial processes, this paper proposes an auxiliary-classifier generative adversarial data augmentation method that integrates local-global interaction and class constraints. Based on the class-conditional generation and auxiliary classification structure of ACGAN, the proposed method sets a temporal convolution branch and a self-attention branch in parallel in the generator to extract local variation patterns and global dependencies within a window, respectively. Bidirectional cross-attention is then used for feature interaction, and a block-level residual connection is introduced. In the training objective, Wasserstein gradient penalty, auxiliary classification supervision, and class-conditional intra-class and inter-class constraints are combined to constrain the class consistency, temporal variation statistics, and variable-correlation structure of generated samples. Supplementary samples are generated according to the sample deficits of each class in the training set, merged with real windows, and then evaluated for operating status by an independent logistic-regression model. Comparative experiments on an imbalanced hydrocarbon-liquid process dataset demonstrate the effectiveness, superiority, and stability of the proposed method under imbalanced conditions.
Downloads
References
[1] Du, Y., Wang, Z., & Wang, X. (2017). Data-driven online operating performance assessment for multi-datasets multivariable industrial processes. Chemical Engineering Transactions, 61, 1729–1734. https://doi.org/10.3303/CET1761286
[2] Fu, W., Wang, Z., & Wang, X. (2025). Industrial process performance assessment method for imbalanced samples based on DVAE-WAFFN-GAN. CIESC Journal, 76(2), 769–786. https://doi.org/10.11949/0438-1157.20240835
[3] Cui, Y., Jia, M., Lin, T. Y., et al. (2019). Class-balanced loss based on effective number of samples. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.
[4] Zhu, J., Wang, Z., Chen, J., et al. (2022). Balanced contrastive learning for long-tailed visual recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 6908–6917).
[5] Du, C., Wang, Y., Song, S., et al. (2024). Probabilistic contrastive learning for long-tailed visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence. https://doi.org/10.1109/TPAMI.2024.3369102
[6] Chawla, N. V., Bowyer, K. W., Hall, L. O., et al. (2002). SMOTE: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321–357. https://doi.org/10.1613/jair.953
[7] Goodfellow, I., Pouget-Abadie, J., Mirza, M., et al. (2014). Generative adversarial nets. In Advances in Neural Information Processing Systems (Vol. 27).
[8] Mirza, M., & Osindero, S. (2014). Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784.
[9] Odena, A., Olah, C., & Shlens, J. (2017). Conditional image synthesis with auxiliary classifier GANs. In Proceedings of the 34th International Conference on Machine Learning (Vol. 70, pp. 2642–2651). PMLR.
[10] Gulrajani, I., Ahmed, F., Arjovsky, M., et al. (2017). Improved training of Wasserstein GANs. In Advances in Neural Information Processing Systems (Vol. 30).
[11] Tian, Y., Shen, J., Wang, A., et al. (2024). Data augmentation and fault diagnosis for imbalanced industrial process data based on residual Wasserstein generative adversarial network with gradient penalty. Journal of Chemometrics, 38(12), e3624. https://doi.org/10.1002/cem.3624
[12] Yoon, J., Jarrett, D., & Van der Schaar, M. (2019). Time-series generative adversarial networks. In Advances in Neural Information Processing Systems (Vol. 32).
[13] Vaswani, A., Shazeer, N., Parmar, N., et al. (2017). Attention is all you need. In Advances in Neural Information Processing Systems (Vol. 30).
[14] Li, X., Metsis, V., Wang, H., et al. (2022). TTS-GAN: A transformer-based time-series generative adversarial network. In Artificial Intelligence in Medicine (Vol. 13263, pp. 133–143). Lecture Notes in Computer Science. https://doi.org/10.1007/978-3-031-09342-5_13
[15] Madane, A., Dilmi, M., Forest, F., et al. (2022). Transformer-based conditional generative adversarial network for multivariate time series generation. arXiv preprint arXiv:2210.02089.
[16] Yue, Z., Wang, Y., Duan, J., et al. (2022). TS2Vec: Towards universal representation of time series. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 36, No. 8, pp. 8980–8987). https://doi.org/10.1609/aaai.v36i8.20881
[17] Woo, G., Liu, C., Sahoo, D., et al. (2022). CoST: Contrastive learning of disentangled seasonal-trend representations for time series forecasting. In International Conference on Learning Representations.
[18] Nie, Y., Nguyen, N. H., Sinthong, P., et al. (2023). A time series is worth 64 words: Long-term forecasting with transformers. In International Conference on Learning Representations.
[19] Wu, H., Hu, T., Liu, Y., et al. (2023). TimesNet: Temporal 2D-variation modeling for general time series analysis. In International Conference on Learning Representations.
[20] Liu, Y., Wijewickrema, S., Li, A., et al. (2024). Time-Transformer: Integrating local and global features for better time series generation. In Proceedings of the 2024 SIAM International Conference on Data Mining (pp. 325–333). SIAM. https://doi.org/10.1137/1.9781611978032.37
[21] Yuan, X., & Qiao, Y. (2024). Diffusion-TS: Interpretable diffusion for general time series generation. In International Conference on Learning Representations.
[22] Coletta, A., Gopalakrishnan, S., Borrajo, D., et al. (2023). On the constrained time-series generation problem. In Advances in Neural Information Processing Systems (Vol. 36).
[23] Narasimhan, S. S., Agarwal, S., Akcin, O., et al. (2024). TimeWeaver: A conditional time series generation model. In Proceedings of the 41st International Conference on Machine Learning (Vol. 235, pp. 37293–37320). PMLR.
[24] Sikder, M. F., Ramachandranpillai, R., & Heintz, F. (2025). TransFusion: Generating long, high fidelity time series using diffusion models with transformers. Machine Learning with Applications, 20, 100652. https://doi.org/10.1016/j.mlwa.2025.100652
[25] Shankar, A., Chen, L. Y., Van Deursen, A., et al. (2025). WaveStitch: Flexible and fast conditional time series generation with diffusion models. Proceedings of the ACM on Management of Data, 3(6), 1–25. https://doi.org/10.1145/3769842
[26] Wen, Y., Zhang, K., Li, Z., et al. (2016). A discriminative feature learning approach for deep face recognition. In Computer Vision — ECCV 2016 (Vol. 9911, pp. 499–515). Lecture Notes in Computer Science. https://doi.org/10.1007/978-3-319-46478-7_31
[27] Khosla, P., Teterwak, P., Wang, C., et al. (2020). Supervised contrastive learning. In Advances in Neural Information Processing Systems (Vol. 33).
[28] Luo, D., & Wang, X. (2024). ModernTCN: A modern pure convolution structure for general time series analysis. In International Conference on Learning Representations.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Academic Journal of Applied Sciences
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
