Date of Award

Fall 2022

Document Type

Open Access Dissertation


Electrical Engineering

First Advisor

Enrico Santi


This work describes novel closed-form, implicitly integrated (CF-implicit) models of switched-mode power converters which feature implicit integration but require no iterative numerical solving algorithm for evaluation because they are explicitly solved prior to model execution. The derived models capture the large-signal dynamic behavior of the power converters, so their use and accuracy are not limited to any one set of operating conditions. These models can be implemented in any computational environment, including directly on an existing embedded controller as a digital twin. Since no iterative solver is required, the models are highly computationally efficient and have a very predictable worst-case execution time, which makes them especially suitable for real-time modeling applications like hardware-in-the-loop simulation or digital twins. The work includes experimental validation in an embedded environment and a discussion of sources of model error. Numerous applications of the modeling technique are explored on a variety of power electronic converters, and computational expense is analyzed. The method is further extended from switch averaged models to switching models and reduced-order models, and the method is qualitatively and quantitatively compared to recently proposed data-driven alternatives based on dynamic neural networks.