A New Energy Management System of Charge Station of Electric Vehicles by Using a Combination Deep Learning Method

The rapid adoption of electric vehicles (EVs) necessitates advanced energy management systems to mitigate grid instability caused by fluctuating charging demand. This study proposes an attention-based Long Short-Term Memory (LSTM) model for predicting EV charging load and optimizing energy allocation. The model leverages historical data from the Adaptive Charging Network (ACN) dataset, incorporating preprocessing techniques such as missing value imputation, feature scaling, and one-hot encoding to enhance data quality. Experimental results demonstrate that the attention-based LSTM outperforms conventional deep learning and machine learning algorithms, achieving a mean squared error (MSE) of 0.0099, mean absolute percentage error (MAPE) of 2.8%, and an accuracy of 98.2%. The model effectively captures temporal dependencies and identifies peak demand periods, enabling efficient integration of renewable energy sources and reducing operational costs. This research highlights the critical role of data preprocessing and advanced deep learning architectures in sustainable energy management for EV charging infrastructure.