Data-driven co-design of adaptive FDI estimation and secure frequency predictive control for networked microgrids under hybrid cyberattacks

This paper investigates the secure control problem of networked wind-energy microgrids subject to hybrid attacks, including sensor false-data-injection (FDI) and aperiodic communication denial-of-service (DoS). To mitigate frequency deviations caused by parameter variations and load fluctuations, we propose a new control-oriented linear data mapping model based on parameter prediction and attack estimation to predict future outputs and control inputs for microgrid frequency regulation. Building on this, an adaptive control algorithm that integrates attack resilience and mitigation is designed to achieve frequency recovery. The main features are as follows: 1) development of a more general FDI attack model that relaxes boundedness constraints, along with a new high-order adaptive observer to estimate FDI attacks; 2) proposal of a novel compensation method that mitigates DoS attack effects by reconstructing hijacked packets using neuro-fuzzy prediction and adaptive estimation; 3) introduction of an online controller gain optimization strategy that integrates pseudo-partial-derivative predictions and attack parameters to enhance system resilience. The boundedness conditions for the frequency and duration of tolerable attacks are rigorously derived. Finally, the proposed strategy is validated through simulations on a dedicated test platform of energy big data.