Weather prediction driven energy management strategy for photovoltaic charging station utilizing electric vehicles as flexible energy storage

The current photovoltaic energy management system mainly adapts passively to the weather conditions and lacks the ability to perceive future weather, which leads to a decline in the economic performance of the system and excessive reliance on the power grid. To address these issues, this paper proposes an energy management and dispatching strategy that adjusts intelligently based on weather prediction information and applies it to a photovoltaic charging station. The framework of the photovoltaic-energy storage charging station system and the charging behavior model for electric vehicle users were established as the foundation. To optimize the system's operation, a hybrid proximal policy optimization algorithm was employed, and a dedicated neural network structure was constructed at the input end to process weather information, thus enhancing the agent's ability to recognize changes in weather types. Furthermore, sensitivity analysis was performed to evaluate the effects of varying weather data accuracies and update frequencies. The results validate the stability and effectiveness of the proposed strategy across various weather scenarios. Compared with the baseline, this strategy increases revenue by 24.1%, reduces the power grid net interaction level by 24.2%, and cuts carbon emissions by 16.4%. This algorithm significantly decreases the frequency of electricity procurement from the grid and enables photovoltaic-energy storage charging stations to shift from passively adapting to the weather to actively utilizing climatic characteristics, enhancing the economy and stability of the system.