Detection and Diagnosis of Long-Term Cyber-Attacks for Predictive Energy Management System in HEVs

In this paper, a systematic cyber-attack detection and diagnosis method is designed to enhance the cybersecurity of hybrid electric vehicles (HEVs). A series of innovative data-driven and model-based residuals are proposed to detect energy-oriented cyber-attacks on the energy management system, which aim to deteriorate the battery capacity and energy efficiency but can hardly be detected by the human driver. Besides the cyber-attacks on critical sensor measurements like vehicle speed, gear position, and battery state of charge, we explore two types of controller attacks, as optimization-based and rule-based damage-oriented controller attacks. Such controller attacks will significantly reduce energy efficiency while satisfying the dynamic performance. Simulation results have shown the effectiveness of the proposed detection method.