Electric Vehicle charging station load forecasting with an integrated DeepBoost approach

The emission of Greenhouse Gases (GHGs) is the main cause of climate change and the transportation sector, especially in urban areas, is among the major contributors to these emissions which are putting a big question mark on sustainability. A significant reduction in emissions is possible through widespread adoption of Electric Vehicles (EVs) which can help in addressing the issues of climate change and sustainability. However, the introduction of EVs brings additional load on the existing grid and can adversely affect its operation. Therefore, we presented a novel DeepBoost approach for forecasting day-ahead EVs charging station load. The proposed approach consists of Categorical Boosting (CatBoost), Extreme Gradient Boosting (XgBoost), Long Short-Term Memory Network (LSTM) and Linear Regression (LR) models. The performance of DeepBoost is compared with conventional CatBoost, XgBoost, LSTM, Informers and different hybrid deep learning methodologies and other techniques reported in the literature. The results demonstrate the effectiveness of the DeepBoost over other models. For the dataset of Adaptive Charging Networks (ACN), the Mean Absolute Error (MAE) of DeepBoost improves by 9.4%, 32.7% and 88% as compared to CatBoost, XgBoost and LSTM networks, respectively. • DeepBoost framework combines parametric and non-parametric techniques effectively. • Robustness analysis of DeepBoost includes SOTA comparison for EV load forecasting. • Literature comparison of DeepBoost shows scalability and generalizability strengths.