AI-Driven Vehicle-to-Grid Decision System for Distributed Energy Storage Management and Grid Resilience

The rapid adoption of electric vehicle (EVs) poses significant challenges to power grid stability, primarily due to their uncoordinated charging patterns. This study proposes a data-driven AI paradigm that optimizes EV charging in urban setting using real-time data and transform EVs into distributed energy storage resources. By training historical charging records and temporal features, the forecast model achieved a strong accuracy ($R 2=0.83$) for electrical demand, enabling effective integration with real time grid demand. The system combines a long short-term memory (LSTM) forecasting model with a threshold-based decision (TBD) strategy to efficiently manage dynamic decisions for EVs in response to real-time grid requirements. The simulation results demonstrate reliable performance in response to grid constraints, highlighting the potential of EVs as distributed grid-friendly energy storage resources. These findings provide valuable information for infrastructure development, particularly for the practical implementation challenges of V2G energy storage management and supporting large-scale deployment through automated decision systems and incentive programs for EV owners.