Smart electric vehicle charging station scheduling with vehicle-to-grid technology

The fair allocation of energy resources at charging stations remains a key challenge, especially during peak demand periods. The inherent flexibility of electric vehicle (EV) loads and their potential to act as distributed energy storage systems offer a viable solution to this problem. However, preserving users' comfort in this setting is complex, as it requires balancing charging needs and battery degradation costs. This study addresses four critical gaps: (i) the role of EVs as batteries in peak shaving, (ii) the station's potential to contribute to grid stability through vehicle-to-grid (V2G) operations, (iii) the importance of fairness in resource allocation to maintain user trust, and (iv) the integration of renewable energy resources with charging stations to minimize environmental pollution. We propose a smart scheduling framework for a local EV charging station with a photovoltaic system to maximize user satisfaction and operational efficiency. The framework controls bidirectional energy and information flow between the station components and users. Charging and discharging tasks are scheduled using a mixed-integer linear programming model that accounts for user preferences, battery degradation costs, and system constraints. The model supports partial fulfillment of charging requests and incentivizes V2G participation while ensuring sufficient charge for onward travel. Simulation results indicate that the proposed framework effectively balances supply and demand, maintains operational stability, and improves user satisfaction through fair energy allocation. Moreover, the model maximizes clean energy and economic V2G utilization with minimum effects on EV owners' mobility. This work demonstrates that EVs can be used as active grid resources and presents a scalable method for integrating distributed energy resources into urban charging infrastructures. As a next step, we plan to enhance the robustness of scheduling with photovoltaic forecasting. To this end, we will explore using long short-term memory networks, which effectively capture temporal patterns and long-term dependencies in nonlinear data.