The rapid proliferation and increasing adoption of electric vehicles (EVs) have rendered them a fundamental component of intelligent transportation networks, contributing significantly to the reduction of harmful greenhouse gas emissions. The surge in the number of EVs necessitates an equally expanding infrastructure to meet their charging requirements. Accurate prediction of EV charging demand, therefore, is critical to alleviate strain on power systems and associated costs. This study presents a novel hybrid deep learning model aimed at predicting the charging needs of electric vehicles. The Convolutional Neural Network (CNN), an integral part of this model, is employed for data collection. The CNN effectively extracts local features of the data, focuses on localized information, and reduces computational demands. The Bidirectional Gated Recurrent Unit (BGRU) contributes to superior performance with time-series data due to its inherent ability to analyze such data. The Empirical Mode Decomposition (EMD) is used to decompose the input time series data while preserving their characteristics. The parameters of the BGRU prediction model are then fine-tuned using a hybrid Jarratt-Butterfly optimization algorithm (JBOA) model. The innovative EMD—CNN—BGRU predictor is evaluated using the EV charging dataset collected from the Georgia Institute of Technology in Atlanta, Georgia, USA. The simulation results achieved an impressive 98% accuracy in prediction. A comparative analysis with existing methods in the literature reveals the superior predictive metrics of the proposed deep learning neural forecaster for the dataset under consideration.