EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

A Multi‐Criteria Optimization Framework for Load Forecasting in Electric Vehicle Charging Stations

Armin Bagherian, Behnam Mohammadi‐Ivatloo, Mehdi Abapour

IET Renewable Power Generation · 2026-01 · 20 (1)

DOI: 10.1049/rpg2.70245 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IET Renewable Power Generation, online erschienen am 03.07.2026. Verfasst von Armin Bagherian und 2 weiteren Autor:innen. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The rapid growth of electric vehicles (EVs) and the construction of charging stations have imposed a lot of stress on distribution networks, highlighting the growing need for accurate forecasting and identifying suitable features for predicting the load of electric vehicle charging stations (EVCSs). This study introduces a data‐driven, optimization‐based decision framework for EVCS load‐forecasting models and feature selection under multiple performance criteria, developed using 6 months of real‐world data from Shenzhen, China. Two new calendar‐based features (public holidays and weekends) were added to the data. The analysis shows that calendar‐based features play an essential role in improving accuracy. To assess the performance, 13 machine learning models and 34 deep learning (DL) models, including basic, combined and hybrid models, were used and compared. Moreover, the first 5 months of data were used to train the model, and the whole sixth month, which the models had not seen, was used for testing. Then, cross‐validation was used on the DL models, and the top 10 models were selected based on this criterion. After the grid search, the stability of the folds was evaluated, and a multi‐criteria mathematical framework with specific constraints was designed to choose the optimal and fair model. This framework was solved using the Mixed‐Integer Linear Programming (MILP) method to determine the best model for this study, considering three main criteria. The results of this study indicate that gated recurrent unit (GRU), bidirectional gated recurrent unit (BiGRU) and temporal convolutional network (TCN) models outperformed other models. Finally, sensitivity analysis on the proposed framework revealed that the suggested method is robust and optimal, providing a reliable and interpretable solution for selecting the appropriate model to predict the load of EVCSs.

Zitieren

APA

Bagherian, A., Mohammadi‐Ivatloo, B., & Abapour, M. (2026). A Multi‐Criteria Optimization Framework for Load Forecasting in Electric Vehicle Charging Stations. IET Renewable Power Generation, 20(1). https://doi.org/10.1049/rpg2.70245

Vancouver

Bagherian A, Mohammadi‐Ivatloo B, Abapour M. A Multi‐Criteria Optimization Framework for Load Forecasting in Electric Vehicle Charging Stations. IET Renewable Power Generation. 2026;20(1). doi:10.1049/rpg2.70245

BibTeX
@article{Bagherian2026,
  author = {Bagherian, Armin and Mohammadi‐Ivatloo, Behnam and Abapour, Mehdi},
  title = {A Multi‐Criteria Optimization Framework for Load Forecasting in Electric Vehicle Charging Stations},
  journal = {IET Renewable Power Generation},
  year = {2026},
  volume = {20},
  number = {1},
  doi = {10.1049/rpg2.70245}
}
Kostenlos starten →