Stochastic optimization of electric vehicle charging strategy based on day-ahead high precision forecast for renewable power and charging demand

Artikel i vetenskaplig tidskrift, 2025

The rapid development of electric vehicles (EVs) will result in significant impacts on the power grid's stability. The uncertainty of renewable power generation and charging demand also leads to the operating costs' increase for the power grid. To address these challenges, this study introduces a multi-objective stochastic optimization model for EV charging regulation based on high-precise forecasts of renewable energy outputs and EV charging demands. The model includes two primary modules: First, a combination of deterministic and probabilistic models is adopted to forecast the renewable power generation and charging demand, with various factors such as temperature, solar irradiation intensity, weather, etc. being considered. The second module generates representative error scenarios to construct the stochastic optimization framework, aiming to minimize load fluctuations and charging costs using the NSGA-II algorithm. Applied to a case study in Shenzhen, the model demonstrates substantial predictive accuracy, with the CNN-LSTM, Bi-LSTM and ETR models achieving R² values of 0.942, 0.978, and 0.954, respectively, for charging demand, wind speed and solar irradiation intensity forecasts. The charging strategy optimization effectively enhanced system efficiency, with increasing the aggregator's profit rate by 30.04 % at most, and reducing power load fluctuations and EV charging costs by up to 18.19 % and 16.57 %, respectively.