New hierarchical control co-design for robust DC microgrids

A team from the University of Notre Dame—Shirantha Welikala, Zihao Song, Hai Lin, and Panos J. Antsaklis—has published a paper on arXiv proposing a unified framework for control and topology co-design in robust networked systems. The paper, submitted to Automatica, introduces two hierarchical design strategies: a model-based approach that assumes full knowledge of subsystem dynamics, and a data-driven approach that relies only on rich input-state-output trajectory data. Both methods leverage dissipativity theory to enforce local and global stability guarantees.

The model-based strategy solves a sequence of linear matrix inequality (LMI) problems, ensuring compositionality and decentralizability without the need for non-convex, centralized iterations. The data-driven variant relaxes the assumption of known dynamics by bounding unknown disturbances with quadratic matrix inequalities and applying the matrix S-lemma. The authors validate their approach on a DC microgrid, demonstrating robust voltage regulation and current sharing. This work is particularly relevant for applications in smart grids, autonomous vehicle platoons, and other large-scale networked systems where both physical topology and control logic must be optimized together.