A GRP hybrid optimization framework with pricing mechanisms for multi-objective EV charging station operation

Abstract

To address the multi-objective optimization of economic performance, grid-load regulation, and environmental performance in electric vehicle (EV) charging-station operation, this paper proposes a Game–Reinforcement Learning–Pareto (GRP) hybrid optimization framework. A Stackelberg game provides the core leader–follower coordination mechanism, while Q -learning and Pareto optimization form two parallel branches for adaptive pricing and economic–environmental balance screening. Seven controlled configurations are constructed to quantify the incremental effects of price response, grid-regulation formulation, dynamic incentives, adaptive pricing, and multi-objective screening under consistent operating conditions. EV behavior is represented by a six-level hierarchical charge-discharge mechanism subject to mobility and battery constraints, together with differentiated incentive mechanisms. Renewable-generation uncertainty is characterized through 500 equally weighted solar–wind scenarios, with a sample-average profile used as a common scheduling input and the full scenario ensemble retained for environmental evaluation. Comparative simulations reveal distinct trade-offs across EV charging cost, operator profitability, grid-load regulation, carbon emissions, and renewable energy (RE) utilization, rather than uniform improvement across all indicators. Supplementary disturbance, initialization, weight-sensitivity, and stochastic-demand analyses are further used to assess the stability and sensitivity of the reported results under the tested conditions. The proposed GRP framework therefore provides a structured basis for quantifying and comparing the contributions and trade-offs of heterogeneous coordination and optimization mechanisms in EV charging-station scheduling.