Optimizing Electric Vehicle Charging Infrastructure through Machine Learning: A Study of Charging Patterns and Energy Consumption

Ayoub Alsarhan; Athari Alnatsheh; Mohammad Aljaidi; Tuqa AL Makkawi; Mahmoud Aljamal; Tamam Alsarhan

doi:10.3991/ijim.v18i21.50843

Authors

Ayoub Alsarhan Department of Information Technology, Faculty of Prince Al-Hussien bin Abdullah for IT The Hashemite University, Zarqa, Jordan https://orcid.org/0000-0001-9075-2828
Athari Alnatsheh Department of Information Technology, Faculty of Prince Al-Hussien bin Abdullah for IT The Hashemite University, Zarqa, Jordan
Mohammad Aljaidi Department of Computer Science, Faculty of Information Technology, Zarqa University, Zarqa, Jordan https://orcid.org/0000-0001-9486-3533
Tuqa AL Makkawi The Hashemite University
Mahmoud Aljamal Department of Information Technology, Faculty of Prince Al-Hussien bin Abdullah for IT The Hashemite University, Zarqa, Jordan https://orcid.org/0009-0007-5389-6778
Tamam Alsarhan King Abdullah II School of Information Technology, The University of Jordan, Amman, Jordan

DOI:

https://doi.org/10.3991/ijim.v18i21.50843

Keywords:

Electric Vehicle (EV), Charging Station Location Analysis, Machine Learning (ML)

Abstract

The rapid adoption of electric vehicles (EVs) has created a pressing need for efficient charging infrastructure. but challenges such as inconsistent demand and poor placement remain. An effective distribution of sufficient EV charging stations (CSs) is one of the major obstacles preventing the market penetration of EVs and the realization of a sustainable transportation system in urban areas. In this paper, a new machine learning technique is proposed in order to optimize the placement of EV charging stations (EVCSs) in metropolitan areas based on an energy consumption prediction model. A dataset from 148,136 charging transactions in Boulder, Colorado, is used with the proposed model. Key algorithms such as KNeighborsRegressor and RandomForestRegressor were incorporated to solve the placement problem. The analysis revealed significant demand fluctuations during peak commute hours, with the KNeighborsRegressor model demonstrating superior prediction accuracy. These insights can guide more effective infrastructure planning and resource allocation, ultimately enhancing the efficiency and user experience of EV charging networks and promoting sustainable urban transportation.