Machine Learning Approach for Predictive Voltage-Constrained Power Capacity in Distribution Network with Electric Vehicles

The rapid adoption of Electric Vehicles (EVs) introduces significant operational challenges to distribution networks, including voltage deviations, peak demand increases, and transformer overloading. This paper proposes a Machine Learning (ML)-based method for predicting voltageconstrained power limits, enabling real-time EV charging control. Two ML models, Linear Regression (LR) and Artificial Neural Network (ANN), were developed and trained using datasets generated from MATLAB Simulink simulations across multiple distribution network topologies. The LR model achieved an accuracy of$\mathbf{R}^{\mathbf{2}} \boldsymbol{>} \mathbf{0. 9 5}$while reducing computation time by approximately 94 % compared to the ANN, confirming its suitability for real-time Energy Management Systems (EMS). Across three representative case studies, the proposed method successfully maintained bus voltages within the$\pm 10 \%$IEC 60038 limits, restoring critical-bus voltage from 204.6 V to 207 V, and correctly identifying the maximum allowable charging power with less than 5 % prediction error. The approach is computationally efficient, data-light, and adaptable to different grid configurations, making it a practical component for integration into optimization-based, voltage-aware EV charging management frameworks.