Abstract
This article presents a novel data-driven methodology designed to enhance the resilience of distributed DC microgrids against various cyber attacks, including Fault Detection and Isolation (FDI) attacks, Denial of Service (DoS) attacks, and delay attacks. A Transformer-based Kalman Filter (TKF) estimator was developed to predict the transmission of signals based on local measurements, addressing the challenges posed by noisy data environments. The proposed approach integrates an AutoRegressive Integrated Moving Average (ARIMA) model to formulate a state-space representation of the microgrid, while leveraging the strengths of deep learning techniques, particularly through the combination of transformers and Long Short-Term Memory (LSTM) networks, for effective high-dimensional data extraction. Extensive simulations conducted in MATLAB and Python demonstrated the efficacy of the TKF estimator in maintaining stable operations of the microgrid under various attack scenarios. The results highlight a significant improvement in estimation accuracy and system performance, validating the robustness of the proposed method. Future research directions are suggested, focusing on the incorporation of advanced filtering techniques and deep learning models to further enhance the system's adaptability and effectiveness in managing nonlinearities and uncertainties in microgrid operations.