Enhancing Efficiency in Megawatt-Scale EV Charging via Adaptive Multi-Agent Control of DC/DC Converters

The rapid adoption of electric vehicles (EVs) demands high-power charging solutions that are efficient, scalable, and reliable. This work introduces a comprehensive simulation framework for megawatt-scale charging systems, focusing on the integration and control of multiple DC/DC converters. With the primary objective of maximizing overall system efficiency during megawatt-scale charging operations. The study addresses key challenges in multi-converter operation, including load distribution, synchronization, and system efficiency. A multi-agent adaptive control strategy was implemented to dynamically optimize operating points and allocate charging currents across converters in real time so that each participating converter operates at its optimal operating point where the maximum efficiency is delivered. This multi-agent adaptive control strategy allocates not only the individual optimal operating points of the multiple DC/DC converters but rather determines the optimal number of participating DC/DC converters at each time instance during the charging process. In addition to that, the strategy provides the option of delivering the optimal charging current during each time instant so that maximum system efficiency is guaranteed during the charging process and reducing overall energy losses. Simulation results demonstrate efficiency improvements of up to 4% compared to conventional static allocation, reducing energy losses and enhancing thermal balance. Additionally, strategies for integrating renewable energy sources and buffer storage were evaluated to further improve system performance. Using the proposed control strategy, a 1 MW charging station can achieve an efficiency increase from 95% to 98%, resulting in annual savings of approximately 120 MWh, €18,000 in energy costs, and an 8 t reduction in CO₂ emissions. For 10 MW installations, these savings scale up to 1.2 GWh per year and 80 t of CO₂. The proposed approach enables intelligent supervisory control for next-generation high-power charging stations, combining efficiency, cost-effectiveness, and sustainability. These findings support the development of modular, resource-efficient infrastructure for future EV ecosystems.