Machine Learning-Driven Demand Prediction for Off-Grid PV-Integrated EV Charging Stations

The rapid growth of electric vehicles (EVs) has led to an increasing demand for charging infrastructure, particularly in off-grid locations. Developing off-grid EV charging stations powered by photovoltaic (PV) systems is crucial to support this expansion. However, the variability of solar power presents challenges in maintaining efficient and reliable charging services. This study focuses on developing off-grid EV charging stations and optimizing load demand prediction using machine learning models to enhance the operational efficiency of these charging stations. Various algorithms, including Random Forest, Decision Tree, SVM, Logistic Regression, and Naïve Bayes, were employed to model and predict load demand based on historical data and real-time inputs. The results indicate that the Random Forest and Decision Tree models achieved the highest accuracy, at 93% and 92%, respectively. The effectiveness of machine learning in dynamically adjusting power distribution according to predicted demand is highlighted, reducing reliance on traditional energy grids and supporting the sustainable development of EV charging solutions in remote areas. Integrating these predictive models allows for better management of energy resources, ensuring a consistent and cost-effective power supply and facilitating the broader adoption of renewable energy in EV infrastructure.