Securing Electric Vehicle Charging Infrastructure: Attack Identification Using Machine Learning

The growing adoption of Electric Vehicles (EVs) is critical for attaining a sustainable energy future, necessitating the development of a reliable Electric Vehicle Charging Infrastructure (EVCI). However, EVCI is vulnerable to cyberattacks, which hinder the rapid adoption of EVs. To mitigate this challenge, this work employs machine learning models to enhance intrusion and anomaly detection in the EV charging ecosystem. Using the recently released EV charger attack dataset CICEVSE2024, we employ machine learning models to identify specific types of cyberattacks on EVCI. A key contribution is the ability to identify not only the presence or category of attacks but also the specific attack types, which can aid in designing attack-resilient EVCI and refining cybersecurity policies. Our findings reveal that machine learning models are effective in securing Electric Vehicle Supply Equipment (EVSE) hardware, achieving over 93% in both accuracy and F1-score in identifying eighteen attack types. The research offers promising insights into strengthening the cybersecurity of EVCI with its increasingly integration with smart grids and smart cities.