Evaluating the prospect of V2G technology in Bangladesh by Optimal Scheduling of Vehicle-to Grid Energy and availability prediction with ML

Abstract

This study examines the prospect of Vehicle-to-Grid (V2G) services in the Mohakhali DOHS region, with a specific emphasis on enhancing grid efficiency by strategically timing electric vehicle (EV) charging and discharging. The work intends to minimize variations in grid demand by utilizing Pyomo and the General Linear Programming Kit (GLPK). This will be achieved by anticipating the availability of electric vehicles (EVs) using machine learning models and constructing an optimization method. The Dhaka Power Distribution Company (DPDC) and the National Travel Survey (UK) provided the data used to train Random Forest and Gradient Boosting models. These models achieved a prediction accuracy of around 90% in determining the availability of electric vehicles (EVs). The optimization model takes into account many sources of revenue, such as fixed energy prices, auxiliary services, and energy sales to the grid. It also includes limitations such as battery capacity and charging rates. The simulations showed considerable potential for generating profits and achieving cost efficiency, even when accounting for the costs associated with battery degradation. The study highlights the advantages of combining machine learning and optimization techniques to improve grid stability and energy management. This work establishes the foundation for adopting Vehicle-to-Grid (V2G) technologies to facilitate sustainable urban energy systems in Bangladesh perspective.