False Data Injection Detection based on Machine Learning Techniques for Connected Electric Vehicles

Abstract

Theglobaltrendintheuseofelectricvehicles(EVs)hastransformed the automotive sector and elevated sustainable transportation to a new level. In this thesis, we examine the changing dynamics of cybersecurity in the global car and charging station industries. As these systems become more interconnected, a serious problem arises in the manipulation of charging station data, which can lead to the creation of false data injection attacks (FDIA). This may trick the system into logging false charging sessions. This research de velops a strong problem description and evaluates the extensive effects of cyberattacks on charging stations in response to this im portant issue, highlighting the critical necessity for an efficient de tection mechanism. In order to strengthen the security of the infrastructure support ing electric vehicle charging, this study introduces sophisticated machine-learningtechniquesforFDIAdetection. Usingbothsuper vised and unsupervised algorithms, the suggested method seeks to detect and resist FDIA attempts. The location of the problem is not clear, whether it is from the client-side, server-side, or external im plementation. Theconstructionofanintrusiondetectionsystemus ing a full dataset is essential for charging station operations, which is demonstrated through simulations and results. Furthermore, a novel artificial neural network-based deep learning method is pre sented. An in-depth comparison examination illuminates the ad vantages and disadvantages of different detection methods. This study makes a substantial contribution to the field of cybersecurity for electric vehicle charging infrastructure