Cyberspace enhancement of electric vehicle charging stations in smart grids based on detection and resilience measures against hybrid cyberattacks: A multi-agent deep reinforcement learning approach

The rapid integration of electric vehicle charging stations (EVCS) into smart grids opens many doors for cybersecurity vulnerabilities, especially hybrid cyberattacks like false data injection (FDI) and denial of service (DoS). This paper addresses the urgent need for a robust framework that enhances the resilience and operational stability of EVCS against these threats. The proposed approach brings together adaptive mechanisms for intrusion detection and mitigation across multiple layers of EVCS with a Hybrid Fuzzy-Multi-Agent Deep Reinforcement Learning (HF-MADRL) strategy. Major highlights are the Gaussian thresholding mechanism design and fuzzy C-Means clustering-based polynomial neurons for anomaly detection and classification of hybrid attacks, respectively. Simulation results on the IEEE 69-bus system show that the proposed HF-MADRL framework guarantees detection accuracy of more than 99 % and under hybrid attacks, the proposed methodology detects and mitigates them in less than 1 s. The voltage and power flow in the EVCS network have become significantly stabilized. Furthermore, the HF-MADRL framework avoids policy degradation in offline training and allows real-time decision-making under dynamic conditions. This work contributes to the securing of EVCS infrastructure through the development of a scalable and adaptive defense mechanism that fosters sustainable and secure smart grid integration for future transportation systems.