ABSTRACT
The widespread availability of electric vehicles has led many to forego gas-powered vehicles. So, to keep up with the rising demand, electric vehicle charging infrastructure, whether public or private, is growing at a rapid pace. The cybersecurity risks linked to EVCS have increased in tandem with their proliferation. With this structure in mind, this piece assesses the potential cyber threats to the EVCS network. The three main parts are training the model, selecting features, and preprocessing. The pre-processing of aberrant data involves cleaning, parsing, and normalizing the data. In this case, the optimal feature selection criterion is mean distortion, which is comprised of peak-to-peak voltage. We trained the model using the KNN-SOM-DT for higher accuracy. The proposed method achieved a higher accuracy of 95.82% compared to existing methods, including KNN and DT.
