Dataset of noise signals generated by smart attackers for disrupting state of health and state of charge estimations of battery energy storage systems

This dataset is generated from real-time simulations conducted in MATLAB/Simscape, focusing on the impact of smart noise signals on battery energy storage systems (BESS). Using Deep Reinforcement Learning (DRL) agent known as Proximal Policy Optimization (PPO), noise signals in the form of subtle millivolt and milliampere variations are strategically created to represent realistic cases of False Data Injection Attacks (FDIA). These signals are designed to disrupt the State of Charge (SoC) and State of Health (SoH) estimation blocks within Unscented Kalman Filters (UKF). The low-magnitude noise signals are specifically crafted to be stealthy, evading easy detection while still effectively causing malfunctions in the estimation processes. Additionally, we introduce a verification case using a different battery model and estimation algorithm to enhance generalization. This case involves high-noise signals with defined thresholds for current and voltage noise levels, which cause significant disruptions to Kalman Filters. These signals serve as a complementary example of adversarial attacks, demonstrating how such noise can destabilize estimation algorithms and lead to critical control errors. This dataset is valuable for researchers and engineers aiming to understand and mitigate the effects of smart cyber-physical attacks on BESS. These attacks can disrupt real-time BESS controllers by injecting false input data related to SoC and SoH, leading to physical control manipulations, increased energy costs, inefficiencies in demand-side management, and incorrect day-ahead scheduling, thereby destabilizing grid operations. The data is reusable in studies focused on enhancing the resilience of SoC and SoH estimation methods, as well as in developing robust defensive strategies against smart DRL-based adversarial attacks.