inttransИнтеллектуальный транспортIntelligent transport3033-6007АО «НИИАС»10.24412/3033-6007-2026-339-53-68inttrans-116Research ArticleИСКУССТВЕННЫЙ ИНТЕЛЛЕКТ И МАШИННОЕ ОБУЧЕНИЕСравнительное исследование методов обнаружения выбросов в данных мониторинга нефтяных скважин на основе SVM и DBSCANComparing SVM and DBSCAN for Outlier Detection in Oil Well Monitoring DataМихайловИ. С.MikhailovI. S.

к. т. н., доцент кафедры прикладной математики и искусственного интеллекта

Ph.D., Associate Professor at the Department of Applied Mathematics and Artificial Intelligence

fr82@mail.ruМьоХлайн ВинMyoHlaing Win

аспирант кафедры прикладной математики и искусственного интеллекта

Postgraduate Student at the Department of Applied Mathematics and Artificial Intelligence

myohlaingwin69287@gmail.comСидоровК. О.SidorovK. O.

аспирант кафедры прикладной математики и искусственного интеллекта

Postgraduate Student at the Department of Applied Mathematics and Artificial Intelligence

kirill.sidoroff2014@yandex.ruНациональный исследовательский университет «Московский энергетический институт» (НИУ МЭИ)РоссияNational Research University «Moscow Power Engineering Institute» (NRU MPEI)Russian Federation202627092026103(39)5368Copyright © Михайлов И.С., Мьо Х., Сидоров К.О., 20262026Михайлов И.С., Мьо Х., Сидоров К.О.Mikhailov I.S., Myo H., Sidorov K.O.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://www.intelligent-transport.ru/jour/article/view/116

Цель исследования – сопоставить результаты One-Class Support Vector Machine (One-Class SVM) и DBSCAN при обнаружении статистических аномалий в данных мониторинга нефтяной скважины и оценить возможность их совместного использования. В анализе использованы шесть параметров многофазного измерителя; после преобразования методом главных компонент (PCA) они представлены тремя главными компонентами, сохраняющими 89,86% общей дисперсии. На выборке из 8639 наблюдений One-Class SVM выявил 424 аномальные точки, DBSCAN – 188; 85 наблюдений обнаружены обоими методами. Для классификации режимов течения по уровню газосодержания SVM с RBF-ядром (C = 100,0, gamma = 1,0) на стратифицированном разбиении 70/30 показал точность 98,26% при F1-мере 0,97–0,99. При 5-кратной стратифицированной кросс-валидации F1_macro составила 0,9853±0,0026. Совместное рассмотрение результатов позволяет выделять согласованные двумя методами, обнаруженные только SVM и обнаруженные только DBSCAN статистические выбросы для последующей экспертной проверки.

The aim of this study is to compare One-Class Support Vector Machine (One-Class SVM) and DBSCAN for detecting statistical anomalies in oil-well monitoring data and to assess their joint use. Six parameters from a multiphase water-cut meter were analyzed; Principal Component Analysis (PCA) reduced them to three principal components preserving 89.86% of the total variance. In a sample of 8639 observations, One-Class SVM identified 424 anomalous points and DBSCAN identified 188, with 85 observations detected by both methods. For flow-regime classification by gas-content level, SVM with an RBF kernel (C = 100.0, gamma = 1.0) achieved 98.26% accuracy with an F1 score of 0.97-0.99 on a stratified 70/30 split. Five-fold stratified cross-validation yielded an F1_macro of 0.9853±0.0026. Joint interpretation separates points detected by both methods from SVM-only and DBSCAN-only detections for subsequent expert review.

обнаружение выбросовметод опорных векторовDBSCANметод главных компонентмониторинг нефтяных скважинмашинное обучениеклассификация режимов теченияаномалииoutlier detectionSupport Vector MachineDBSCANPrincipal Component Analysisoil well monitoringmachine learningflow regime classificationanomaliesИсследование было проведено при поддержке РНФ (проект № 24-11- 00285).The study was conducted with the support of the Russian Science Foundation (project No. 24-11-00285).ReferencesEbrahimi, M. Anomaly detection in oil-producing wells: a comparative study of one-class classifiers in a multivariate time series dataset / M. Ebrahimi, E. Khamehchi, M. Ameri // Journal of Petroleum Exploration and Production Technology. – 2023. – Vol. 13. – P. 1845–1860.Ebrahimi, M., Khamehchi, E., & Ameri, M. (2023). Anomaly detection in oil-producing wells: A comparative study of one-class classifiers in a multivariate time series dataset. Journal of Petroleum Exploration and Production Technology, 13, 1845–1860.Mohammadpour, A. Outlier detection and selection of representative fluid samples using machine learning: a case study of Iranian oil fields / A. Mohammadpour, D. Shahsavani, M. Riazi // Journal of Petroleum Exploration and Production Technology. – 2024. – Vol. 14, No. 10. – P. 2833–2852.Mohammadpour, A., Shahsavani, D., & Riazi, M. (2024). Outlier detection and selection of representative fluid samples using machine learning: A case study of Iranian oil fields. Journal of Petroleum Exploration and Production Technology, 14(10), 2833–2852.Liu, Y. Identification Method of Tubing Leakage Based on PCA-SVM in Gas Well / Y. Liu, H. Wang, Z. Chen // Proceedings of the 2nd International Conference on Signal Processing and Intelligent Computing (SPIC). – 2024. – P. 1–6.Liu, Y., Wang, H., & Chen, Z. (2024). Identification method of tubing leakage based on PCA-SVM in gas well. In Proceedings of the 2nd International Conference on Signal Processing and Intelligent Computing (SPIC) (pp. 1–6).Wang, L. Identification method for true and false anomalies of oil pipeline leakage in an old industrial oil production well site / L. Wang, X. Zhang, J. Liu // Oil & Gas Storage and Transportation. – 2023. – Vol. 42, No. 5. – P. 567–575.Wang, L., Zhang, X., & Liu, J. (2023). Identification method for true and false anomalies of oil pipeline leakage in an old industrial oil production well site. Oil & Gas Storage and Transportation, 42(5), 567–575.Comparison of Isolation Forest and One Class SVM in Anomaly Detection of Gas Pipeline Operation // Proceedings of the 2023 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS). – IEEE, 2024. – (IEEE Xplore, 13 February 2024).Comparison of isolation forest and one class SVM in anomaly detection of gas pipeline operation. (2024). In Proceedings of the 2023 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS). IEEE.Reconstruction of well logging data using unsupervised machine learning based outlier detection techniques (UML ODTs) under adverse drilling conditions / J.-R. Chen, R.-Z. Yang, T.-T. Li [et al.] // Applied Geophysics. – 2025. – DOI 10.1007/s11770-025-1178-z.Chen, J.-R., Yang, R.-Z., Li, T.-T., Xu, Y.-D., & Sun, Z.-P. (2025). Reconstruction of well logging data using unsupervised machine learning based outlier detection techniques (UML ODTs) under adverse drilling conditions. Applied Geophysics. https://doi.org/10.1007/s1-025-1178-zLundberg, S. M. A Unified Approach to Interpreting Model Predictions / S. M. Lundberg, S.-I. Lee // Advances in Neural Information Processing Systems. – 2017. – Vol. 30. – P. 4765–4774.Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.Ahmadi, M. Data-Driven Modeling of Oil Saturation from Dielectric Logs Using Ensemble Regression, Dimensionality Reduction and Anomaly Detection Machine Learning Algorithms / M. Ahmadi, M. Bagheri, A. Sadeghi // SPE Annual Technical Conference and Exhibition. – 2023. – Paper SPE-215078-MS.Ahmadi, M., Bagheri, M., & Sadeghi, A. (2023). Data-driven modeling of oil saturation from dielectric logs using ensemble regression, dimensionality reduction and anomaly detection machine learning algorithms. In SPE Annual Technical Conference and Exhibition (Paper SPE-215078-MS).Patel, D. Oil and gas flow anomaly detection on offshore naturally flowing wells using deep neural networks / D. Patel, M. Shah, K. Patel // Journal of Petroleum Science and Engineering. – 2024. – Vol. 234. – Art. 109876.Patel, D., Shah, M., & Patel, K. (2024). Oil and gas flow anomaly detection on offshore naturally flowing wells using deep neural networks. Journal of Petroleum Science and Engineering, , Article 109876.Hassan, N. Unsupervised machine learning model for predicting anomalies in subsurface safety valves and application in offshore wells during oil production / N. Hassan, A. El-Sayed, S. Ibrahim // Journal of Petroleum Exploration and Production Technology. – 2023. – Vol. 13. – P. 2211–2225.Hassan, N., El-Sayed, A., & Ibrahim, S. (2023). Unsupervised machine learning model for predicting anomalies in subsurface safety valves and application in offshore wells during oil production. Journal of Petroleum Exploration and Production Technology, 13, 2211–2225.Zhao, Q. Varying-scale HCA-DBSCAN-based anomaly detection method for multi-dimensional energy data in steel industry / Q. Zhao, M. Li, Y. Sun // Information Sciences. – 2023. – Vol. 645. – Art. 119479.Zhao, Q., Li, M., & Sun, Y. (2023). Varying-scale HCA-DBSCAN-based anomaly detection method for multi-dimensional energy data in steel industry. Information Sciences, 645, Article

The authors declare that there are no conflicts of interest present.