Resilient Consensus-Based Distributed Dispatch over a Network Using Digital-Twin Backup System under Multichannel DoS Attacks

This article addresses the consensus-based distributed dispatch problem for multiple generators by incorporating a backup system for dealing with denial-of-service (DoS) attacks. A novel backup system-based approach has been proposed, motivated by digital twins, by considering generator information and incorporating predictive neural networks (NNs) under the occurrence of DoS attacks on a smart grid (SG). This approach integrates the machine learning predictive feature of XGBoost with a consensus protocol to mitigate the multichannel DoS attacks. The convergence and stability analysis has been provided for the distributed optimization by the application of multiple Lyapunov functions, Lyapunov redesign, and consideration of the absence or occurrence of attacks. The predictive capability of the proposed method allows the dispatch framework to intelligently forecast, adjust, and allocate energy based on dynamic inputs, such as demand fluctuations or DoS-induced disruptions. Compared with conventional methods addressing DoS attacks, this approach is superior: 1) in dealing with DoS attacks due to the consideration of an offline trained dynamic backup system, and 2) can handle multichannel DoS attacks due to a digital-twin-oriented backup system for predicting the generator behavior. Finally, an extensive simulation study is presented to model the NNs, analyze different predictive NNs, and deploy NNs for DoS mitigation.