False Data Injection Attack Detection in EV Charging Network Using NARX Neural Network

Abstract

Cyber threats in electric vehicle (EV) charging networks have become prevalent and target vehicle’s charging processes and power supply from the grid. It has been observed that the existing state-of-the-art schemes for false data injection (FDI) attack detection cannot detect the injection, which is random or unpredictable in nature and persists for short time intervals. Most importantly, it remains undetected when the impact of injection mimics the natural behavior of the EV charging process. Therefore, in this article, a deep learning (DL)-based FDI attack detection scheme is proposed for the EV charging network. The nonlinear autoregressive exogenous (NARX) input neural network (NN) is used to estimate the energy (kWh) delivered to an EV during its charging session. The error of estimation (EoE) obtained from the sensed and estimated values is further analyzed using the interquartile range (IQR) technique, and the attack is detected by identifying a few consecutive spikes given by IQR. The proposed attack detection method is evaluated using a real-world EV charging dataset and compared with the existing state-of-the-art attack detection scheme. The simulation results indicate that the proposed attack detection scheme outperforms the other schemes by achieving an attack detection accuracy of 99.40%, whereas the existing schemes give 88.68% and 98% accuracies, respectively.