Subsea Valve Anomaly Detection: A Comparative Analysis of One-Class Deep Neural Network Architectures

• Lesly Shessira Ttito Ugarte UFF
• Flavia Bernardini UFF
• Eduardo Andrade UFF
• Felipe Murad UFF
• Esteban Gonzalez Clua UFF

DOI: https://doi.org/10.5753/eniac.2026.30569

Resumo

The subsea valves used in oil and gas extraction processes operate under extreme conditions, and anticipating their anomalies raises important social, technical, and economic implications. This paper proposes a framework based on the Deep Neural Networks for One-Class Classification paradigm on multivariate time series to detect rare anomalies. Based on literature reviews, we designed our framework and observed that our adopted paradigm was not widely adopted, as only one model type is typically used. We evaluate three DNN architectures (CNNs, LSTMs, and Informer) on the public 3W dataset. The empirical results show that the Informer architecture achieves the best results in 4 out of 6 scenarios and the second-best in the other 2.

Palavras-chave: Anomaly Detection, Subsea Valve, Deep Neural Network, One Class Classifiers, Time Series

Referências

An, Z., Cheng, L., Guo, Y., Ren, M., Feng, W., Sun, B., Ling, J., Chen, H., Chen, W., Luo, Y., and Yang, Z. (2022). A novel principal component analysis-informer model for fault prediction of nuclear valves. Machines, 10(4).

Aranha, P. E., Lopes, L. G. O., Paranhos Sobrinho, E. S., Oliveira, I. M. N., de Araújo, J. P. N., Santos, B. B., Lima Junior, E. T., da Silva, T. B., Vieira, T. M. A., Lira, W. W. M., Policarpo, N. A., and Sampaio, M. A. (2024). A system to detect oilwell anomalies using deep learning and decision diagram dual approach. SPE Journal, 29(03).

Griffiths, I. (2021). Automated condition monitoring using artificial intelligence. In Proc. 2020 4th Int. Conf. Vision, Image and Signal Processing, ICVISP 2020. ACM.

Harrouz, A., Toubakh, H., Mohamed Redouane, K., Sayed Mouchaweh, M., and Salem, H. (2022). Self adaptive learning scheme for fault prognosis in oil wells and production and service lines. Annual Conference of the PHM Society, 14.

Hausler, R., Krishnamurthy, R., and Sherar, B. W. (2015). Observation of productivity loss in large oil wells due to scale formation without apparent production of formation brine. In NACE CORROSION, pages NACE–2015. NACE.

Hochreiter, S. and Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8):1735–1780.

Khan, S. and Madden, M. (2014). One-class classification: Taxonomy of study and review of techniques. The Knowledge Engineering Review.

LeCun, Y., Boser, B., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W., and Jackel, L. D. (1989). Backpropagation applied to handwritten zip code recognition. Neural computation, 1(4):541–551.

Lian, Y., Geng, Y., and Tian, T. (2023). Anomaly detection method for multivariate time series data of oil and gas stations based on digital twin and MTAD-GAN. Applied Sciences, 13(3).

Machado, A. P. F., Munaro, C. J., and Ciarelli, P. M. (2025). Enhancing one-class classifiers performance in multivariate time series through dynamic clustering: A case study on hydraulic system fault detection. Expert Systems with Applications, 286.

Theyab, M. and Theyab, M. A. (2018). Severe slugging control: Simulation of real case study. J Environ Res, 2(1):5.

Vargas, R. E. V., Munaro, C. J., Ciarelli, P. M., Medeiros, A. G., do Amaral, B. G., Barrionuevo, D. C., de Araújo, J. C. D., Ribeiro, J. L., and Magalhães, L. P. (2019). A realistic and public dataset with rare undesirable real events in oil wells. Journal of Petroleum Science and Engineering, 181:106223.

Zhao, Q., Cheng, G., Han, X., Liang, D., and Wang, X. (2021). Fault diagnosis of main pump in converter station based on deep neural network. Symmetry, 13(7).

Zhou, H., Zhang, S., Peng, J., Zhang, S., Li, J., Xiong, H., and Zhang, W. (2021). Informer: Beyond efficient transformer for long sequence time-series forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35.