#defense #resilience #power-systems #microgrids

As microgrids (MGs) play an increasingly vital role in modern power systems, ensuring their resilience against cyber threats has become a pressing challenge. This talk presents two complementary approaches to ensuring the secure and reliable operation of MGs under such adversarial conditions. The first approach focuses on model-based cyber-resilient control, where advanced observer designs, such as sliding-mode and projection-operator-based unknown input observers, enable accurate attack reconstruction and robust control against data manipulation. The second approach introduces a data-driven intelligent cybersecurity framework, leveraging recurrent neural networks with Laplace prior Bayesian inference to achieve fast, adaptive detection of multiple attack vectors. An extended generative mechanism is further developed to handle impulsive noise and unknown input patterns, enhancing robustness in real-time operations. Validations through OPAL-RT and MATLAB simulations demonstrate the efficacy of these frameworks in mitigating the impact of complex cyber threats, achieving faster dynamic response and improved resilience compared to conventional methods.