A blockchain-integrated, energy-efficient dual-agent reinforcement learning framework for resilient electric vehicles

Abstract

This paper introduces an energy-efficient, Blockchain-Based, Dual-Agent, Reinforcement Learning (BDARL) model of resilient EV-grid integrative effect. All EV energy transactions and agent decisions are validated by the blockchain layer that is integrated through a lightweight proof-of-stake consensus mechanism, and this ensures the tamper-proof functioning and decentralized trust. The proposed framework is applying to a MATLAB/Simulink-Python TensorFlow-Hyperledger Fabric co-simulation environment where the performance analysis shows a 98.8 per cent Resilience Coordination Index (RCI), a 36.7 per cent Energy Efficiency Gain (EEG), a 31.5 per cent Load Stabilization Score (LSS), and a transaction latency of 25 ms. Compared to baseline DRL and non-blockchain schedulers, BDARL offers 7.9% improvement in terms of resilience, 8.4% in terms of energy efficiency, and 14 ms better convergence, so it provides a safe, sustainable, and smart paradigm of managing next-generation EV-grid synergy.