False Data Injection Attack Against State-of-health Estimation of Lithium Battery with Data Generation

Lithium-ion batteries (Libs), as the energy backbone of drones, electric vehicles, and energy storage stations, are vulnerable to false data injection attacks (FDIA) in the network environment. This paper focuses on the study of false data generation to implement the FDIA regarding the data-driven state of health (SOH) estimation process of Libs. Leveraging the powerful nonlinear mapping and automatic feature learning capabilities of machine learning, our proposed model is capable of generating carefully designed FDIA false data samples with any given battery capcity. These samples can strategically launch FDIA on Libs SOH estimation, driving the estimation results to align with our desired attack targets and improving the efficiency of FDIA. To validate the effectiveness of the model, we conducted experiments on the electric vertical takeoff and landing (eVTOL) battery dataset. Results show that the samples generated by the proposed model can accurately achieve the attack targets, misleading the Libs SOH estimation results.