Abstract
The impact on the power grid can be reduced by using smart charging systems to manage the charging of electric vehicles in a more organized and effective manner. Demand response programs, where charging stations may connect with the power grid and modify their charging rate according on the demand for electricity at that moment, are one example of a smart charging solution. This may reduce the need for expensive infrastructure upgrades and balance the strain on the electricity system. In order to balance the electricity grid, vehicle-to-grid (V2G) technology can potentially be deployed. The difficulties posed by electric vehicle charging stations are being addressed through the implementation of smart charging solutions and machine learning techniques. A possible strategy to reduce voltage fluctuation, load variance, and improve power quality is predicting when to offer rapid charging depending on power demand. Finding the best model that can accurately predict the fast-charging times can be done by using ML algorithms like Kernel-Support Vector Machine, Random Forest Classifier, K-Nearest Neighbors, Decision Tree Classifier, Naive Bayes, Artificial Neural Network, Long Short-Term Memory, XGBoost, and CatBoost. The three most effective ML models were picked for this paper: Random Forest Classifier, Decision Tree Classifier, and XGBoost. The top model, XGBoost, known for its enhanced performance, was chosen based on the findings.