• 2025-04-25
• T.O. Ajewole, G.O. Adeyemo, O. Oladepo, A.A. Olawuyi, K.A. Hassan
• Volume 1, Issue 1

Abstract

This study proposes an advanced energy theft detection system using machine learning models, namely Firefly Algorithm-CatBoost (FA-CatBoost) and Genetic Algorithm-CatBoost (GA-CatBoost), to accurately identify fraudulent energy consumption patterns in power grids. Daily power consumption dataset was employed to train and evaluate the models. The Models were evaluated using metrics performance metrics like accuracy, precision, recall, F1-score, Matthews Correlation Coefficient (MCC), and Receiver Operating Characteristic Area Under the Curve (ROC-AUC). The FA-CatBoost model demonstrated exceptional performance, achieving an accuracy of 97.6%, precision of 97.3%, recall of 98.0%, F1-score of 97.7%, MCC of 95.3%, and ROC-AUC of 97.6% during training. The GA-CatBoost model also exhibited promising results with an accuracy of 84.5%, precision of 86.0%, recall of 82.4%, F1-score of 84.2%, MCC of 69.1%, and ROC-AUC of 84.5% during training. The proposed energy theft detection system showcases the effectiveness of advanced machine learning techniques in identifying fraudulent activities within power grids, contributing to the enhancement of energy security and grid resilience. The findings suggest that power utility companies should consider integrating these validated models into their fraud detection systems and invest in data analytics infrastructure to combat energy theft effectively.