Hierarchical decentralized optimal control and reconfiguration of networked microgrids in the power distribution system

1 online resource (160 pages) : PDF

2023

University of North Carolina at Charlotte

Advancements in information and communication technology, decentralized digital economic structures, and data-driven learning-based technology have transformed distribution networks as a system of systems in recent decades. With the rapid integration of sustainable energy resources, the ecosystem surrounding the electricity consumer is getting socially, economically, and politically complex. The key operational challenge is the coexistence of large-scale distributed energy resources (DER) to achieve stable load power sharing while regulating the voltage and frequency in the network to the nominal values. The vision of the dissertation work is to formulate a hierarchical decentralized control structure to accommodate three-level research objectives. First, at the DER level, considering the low X/R and unbalanced nature of the distribution network, appropriate cascaded primary control loops are designed. A unified control architecture is proposed for stable multiple DER power sharing, achieving ride-through capability, and maintaining the network voltage and frequency close to nominal values. The unified control architecture is devised through a systematic definition of steady-state operating modes and the interaction among hierarchical entities in the grid. Second, at the microgrid level, a decentralized predictive optimal constrained secondary control framework to maintain the nominal voltage and frequency is formulated. The proposed strategy is built on a first-order model of the primary controller and local/global measurements-based state estimation, facilitating the deployability to grid edge devices. The framework is further extended to incorporate a data-driven approach when model parameters are unavailable. Finally, at the network level, detailed network dynamics are modeled in a real-time environment by incorporating primary, and secondary control and protection functions. The reinforcement learning agent is designed by utilizing an extended Q-routing methodology, which interacts with the environment through event-driven communication and performs optimal network reconfiguration during events in the environment. Another goal of this dissertation work is to bring value to engaged stakeholders in the process of achieving a 100% sustainable power grid. There exists an execution gap between the aforementioned hierarchical technology solutions and business delivery models. This gap is addressed in the dissertation by fostering an implementation strategy for resiliency services through the energy-as-a-service model. The regulatory framework and ownership agreements are yet to evolve to support the delivery model acceptable to the involved stakeholders.

doctoral dissertations

Electrical engineering

Ph.D.

Grid Forming Inverter
Hierarchical Control
Microgrid
Model Predictive Control
Reinforcement Learning
State Estimation

Electrical Engineering

Kamalasadan, Sukumar

Cecchi, Valentina
Ebong, Abasifreke
Keanini, Russel

Thesis (Ph.D.)--University of North Carolina at Charlotte, 2023.

This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s). For additional information, see http://rightsstatements.org/page/InC/1.0/.

Copyright is held by the author unless otherwise indicated.

Ingalalli_uncc_0694D_13326

http://hdl.handle.net/20.500.13093/etd:3050