| GridShield: A Cyber-Secure AI Framework for Real-Time Detection of Electricity Theft Based on Daily Consumption Behavior |
| Paper ID : 1107-ICEEM2025 (R1) |
| Authors |
|
Ahmed Ramadan *1, Marwa A. Shouman2, Gamal Attiya3, A. S. Zein El Din4, Elhossiny Ibrahim5 1Electrical Engineering Department., Faculty of Enginerring,Menoufia University Shebin Elkom,32511,Egypt 2Computer Science and Engineering Department Faculty of Electronic Engineering, Menoufia University Menouf, Egypt 3Computer Science and Engineering Department Faculty of Electronic Engineering, Menoufia University Menouf, Egypt 4Electrical Engineering Department, Faculty of Engineering, Menoufia University Shebin Elkom, Egypt 5Department of Computer Science and Engineering Faculty of Electronic Engineering, Menoufia University Menouf, Egypt |
| Abstract |
| Electricity theft is a formidable problem facing electricity utility operators which causes immense non-technical losses of at least US96 billion per year. Due to popularization of smart grids and the Advanced Metering Infrastructure (AMI), the power distribution grid has been made both better and smarter yet more exposed to advanced cyber-physical threats like false data injection attacks, spoofing, and malicious access This paper introduces an end-to-end system that combines machine learning-based detection engine and cybersecurity mechanisms to detect and prevent electricity thefts in real-time. We adopt a behavior-aware 24-hour segmentation approach to obtain daily consumption habits, and a stacking ensemble that integrates Random Forest, XGBoost and LSTM models. Moreover, it is enhanced with AES-256 encryption, mutual authentication based on public key infrastructure (PKI), the hybrid intrusion detection system (IDS), and logging using blockchain. The results of experiments on the SGCC dataset indicate a detection F1-score of 96.2% and AUC-ROC = 0.981, with a success rate of the mitigation of cyberattack equal to 97%, which illustrates the promise of the proposed solution as a means of establishing a secure and intelligent smart grid environment. |
| Keywords |
| smart grid, blockchain, LSTM, XGBoost, stacking ensemble, SGCC. |
| Status: Accepted |