Digital Twin-Based Cooperative Control Techniques for Secure and Intelligent Operation of Distributed Microgrids

Networked microgrids play a key role in constructing future active distribution networks for providing the power system with resiliency and reliability against catastrophic physical and cyber incidents. Motivated by the increasing penetration of renewable resources and energy storage systems in the distribution grids, utility companies are encouraged to unleash the capabilities of the distributed microgrid to work as virtual power plants that can support the power systems. The microgrids nature is transforming the grid and their control systems from centralized architecture into distributed architectures. The distributed networked microgrids introduced many benefits to the future smart grids, it created many challenges such as the absence of centric oversight, the lack of robustness against renewable uncertainty and vulnerability to cyberattacks. These challenges and issues imposed the necessities to transform the control system architecture form classical decision making to holistic, hierarchical multi-vision distributed smart decision making. In this dissertation, a data-centric oversight layer is added on top of the energy cyber- physical system to collect the required information by the internet of things (IoT) technologies and provision it to the cloud virtual space. This layer has relatively unlimited computational and communication resources that can guide the distributed cyber system to secure, reliable, efficient and intelligent operation. By adding the data-informed model philosophy (Digital Twin) into this centric layer can answer real-time What-IF questions, enhance the control system resiliency, provide the control system with guidance toward the global optimal objective and provide situational and security awareness. The dissertation extended the developed control system resiliency by providing multi- mode of operation according to the cyber system state. The distributed multi-agent control system is designed to be security-aware without the centric oversight via developing distributed cyberattack observer that can detect and identify the attacked neighbors by extracting the cyber graph and the consensus control features and compare it with the healthy characteristics. Finally, the controller itself is armed with an independent mode of operation, which makes the controller work with the local information only to guarantee stable and optimal operation. The developed techniques and ideas we experimentally tested and evaluated on the FIU smart grid testbed