APPROCHE HYBRIDE COMBINANT LA BLOCKCHAIN ET L’INTELLIGENCE ARTIFICIELLE POUR LA PROTECTION DES INFRASTRUCTURES CRITIQUES DE SOCIÉTÉ NATIONALE D’ELECTRICITÉ(SNEL)

Abstract

Electrical infrastructures are critical systems whose service continuity depends on the rapid detection of anomalies and the reliability of supervisory information. In the context of smart grids, conventional monitoring systems based on SCADA exhibit limitations in terms of automated data analysis and event traceability. Furthermore, machine learning techniques improve fault detection in electrical networks but do not always guarantee the integrity or auditability of the decisions produced. This article proposes a hybrid approach combining machine learning and blockchain to enhance both anomaly detection and the security of alerts within an electrical network. Three-phase electrical signals (currents and voltages) are analyzed using several supervised classification models, including K-Nearest Neighbors, Support Vector Machine, and Random Forest. An experimental evaluation conducted on a publicly available dataset (Electrical Fault Detection and Classification, Kaggle) comprising 12,001 observations simulating three-phase electrical signals shows that the Random Forest model provides the best performance in terms of precision, recall, and F1-score. The alerts generated by the model are then transformed into event reports and recorded in a permissioned blockchain secured by SHA-256 cryptographic hashing functions, ensuring data integrity and event traceability. The results demonstrate that integrating machine learning and blockchain enhances the reliability of decision-making and contributes to the protection of critical infrastructures in smart grids.