BTCN-Based Distributed Resilient Control for Hybrid AC/DC Microgrids With Missing Data

Data-driven methods have been widely applied in distributed microgrids (MGs) to defend against cyber attacks. However, in practical applications, issues such as communication failures and PMU malfunctions often result in incomplete or even unavailable training data. This paper investigates the distributed resilient secondary control problem for hybrid AC/DC MGs under hybrid false data injection (FDI) and denial-of-service (DoS) attacks with missing data. Specifically, the generative adversarial network (GAN), an emerging unsupervised deep learning technique based on two contesting deep neural networks, is employed to generate the missing data. Additionally, a Bayesian temporal convolutional network (BTCN)-based estimator is developed to infer aggregated communication signals for resilient control, with cyber-attack mitigation and system stability ensured by Lyapunov theory. Compared with traditional data-driven resilient control approaches, the proposed method enables robust defense against hybrid attacks under missing data conditions, with improved uncertainty estimation and generalization capabilities provided by the Bayesian framework. Finally, the effectiveness of the proposed method under hybrid attacks is confirmed through a real-time experiment in OPAL-RT.