Pricing Strategy for V2G Based on Charging Behavior Forecasting and Endogenous Carbon Costs

In the context of electric vehicle-grid interaction, this study integrates graph convolutional networks and the Transformer architecture model to predict user charging and discharging behaviors, constructs a Stackelberg game-based pricing model, and incorporates carbon costs into electricity price decision-making. A dynamic pricing strategy is proposed, which appropriately sets charging pile electricity prices and discharge subsidies to optimize users′ charging time choices and incentivize them to feed power back to the grid during peak hours. Experiments show that the model significantly increases the operator′s average daily revenue, effectively reduces carbon emissions, and maintains revenue stability amid fluctuations in carbon prices and power supply costs, thereby achieving the dual objectives of peak shaving and carbon reduction. This provides a valuable reference for low-carbon operation in V2G environments.