Dynamic Load Management System for EV chargers
Mamin, Ildar (2026)
2026
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202602132831
https://urn.fi/URN:NBN:fi:amk-202602132831
Tiivistelmä
This bachelor’s thesis develops a smart dynamic load management (DLM) system for electric vehicle (EV) charging, commissioned by Digi Energia Oy (Ipark). The project addresses the need for cost-effective solutions to prevent grid overloads and improve operational efficiency.
The development focused on a predictive load-balancing system that dynamically allocates power across multiple stations. Key components include a Machine Learning model for peak-demand forecasting, a real-time monitoring dashboard, and adaptive allocation strategies compliant with OCPP standards. Utilizing a quantitative methodology, the system was tested using simulated sensor data and historical charging records.
Results indicate that predictive DLM significantly reduces peak demand and enhances system stability without requiring infrastructure upgrades. The findings highlight improved energy utilization and increased visibility via software-driven optimization. Recommendations include deploying these predictive models across wider networks and conducting further testing under varied operational scenarios.
The development focused on a predictive load-balancing system that dynamically allocates power across multiple stations. Key components include a Machine Learning model for peak-demand forecasting, a real-time monitoring dashboard, and adaptive allocation strategies compliant with OCPP standards. Utilizing a quantitative methodology, the system was tested using simulated sensor data and historical charging records.
Results indicate that predictive DLM significantly reduces peak demand and enhances system stability without requiring infrastructure upgrades. The findings highlight improved energy utilization and increased visibility via software-driven optimization. Recommendations include deploying these predictive models across wider networks and conducting further testing under varied operational scenarios.