Smart Battery Management in Smart Grids and EVs: A Game Theory-Based Approach for Cyber security and Intrusion Detection

Abstract

The convergence of smart grids and Electric Vehicles (EVs) has expanded the role of smart battery systems beyond conventional energy storage, introducing a new set of cyber security challenges. These systems, which form the backbone of modern power infrastructure, are increasingly reliant on IoT-based communications and bidirectional energy exchange mechanisms such as Vehicle-to-Grid (V2G). This paper proposes a game theory-based hybrid machine learning framework to enhance cyber security and intrusion detection in Smart Battery Management Systems (SBMS) deployed across both smart grids and EV ecosystems. A Nash Equilibrium-based game-theoretic model is developed to optimize the allocation of defensive resources against strategic cyber adversaries. This is combined with a hybrid machine learning approach that integrates Support Vector Machines (SVM) and Auto encoders, achieving a detection accuracy of 92.3% and reducing the false positive rate to 5.1%, outperforming traditional models like Random Forest and LSTM. Validation is performed using case studies from Indian smart grid projects, including EV charging infrastructures. The model successfully detects multiple real-world threats, including billing fraud and malware attacks, yielding cost savings of INR 2.3 crore annually. The research aligns with India's National Cyber security Policy 2020 and offers practical insights for securing future energy and mobility infrastructures.