MULTILEVEL VERIFICATION SYSTEM FOR MITIGATING CYBER RISKS IN UPSTREAM OPERATIONS OF SMART OILFIELDS

Abstract

The increasing digitalization of upstream oil and gas operations into "smart oilfields" has introduced significant cyber vulnerabilities, necessitating advanced mitigation strategies. This research proposes a multilevel verification system, integrating a Long Short-Term Memory (LSTM) deep learning model for real-time cyberattack detection. Utilizing synthetically generated multisource data (sensor readings, network traffic, system logs, and device metadata), the system employs rigorous preprocessing, feature engineering, and temporal splitting to prevent data leakage and address class imbalance via SMOTE. The LSTM model achieves near-perfect performance, with 99.71% testing accuracy, 100% recall, and zero false negatives on an independent test set. Simulated real-time monitoring demonstrates high-confidence alerts and conceptual automated responses, enhancing operational resilience. This framework provides a robust blueprint for securing critical energy infrastructure against evolving threats, particularly in oil-producing nations like Nigeria.