Commitment-driven penalty mechanism with dynamic pricing for V2V energy trading: A multi-armed bandit reinforcement learning and game theoretic approach

Electric vehicles (EVs) are rapidly emerging as the optimal alternative to internal combustion engine vehicles, offering a greener and more eco-friendly mode of transportation. While the rapid adoption of EVs poses significant challenges to the electrical grid, vehicle-to-vehicle (V2V) charging presents a feasible solution to alleviate these issues. A significant challenge in the widespread adoption of V2V energy trading is developing an effective V2V energy management framework that maximizes participant benefits. This paper introduces a penalty based holistic energy management framework for V2V energy trading, incorporating a multi-armed bandit reinforcement learning based dynamic pricing strategy to enhance financial benefits for both energy demanding and energy supplying EVs. The penalty mechanism addresses the issue of EVs refusing to charge or discharge based on the outcome of the V2V matching protocol. A real-world data is used in the simulation to offer a practical evaluation of the proposed approach. Simulation results validate the effectiveness of the proposed V2V energy management framework in enhancing EV utility and matching optimality. Comparisons between the approach with and without penalties show that achieving efficient, practical, and beneficial V2V matches is possible, which supports the potential adoption of this framework.