Graduation Year
2024
Document Type
Dissertation
Degree
Ph.D.
Degree Name
Doctor of Philosophy (Ph.D.)
Degree Granting Department
Mechanical Engineering
Major Professor
Tansel Yucelen, Ph.D.
Committee Member
Mehrdad Pakmehr, Ph.D.
Committee Member
Rajiv Dubey, Ph.D.
Committee Member
Kyle Reed, Ph.D.
Committee Member
Sriram Chellappan, Ph.D.
Keywords
Gain Scheduling, Model Reference Adaptive Control, Nonlinear Stability and Control, Uncertain Dynamical Systems
Abstract
Feedback control architectures are fundamental for achieving a desired response in physical systems. A robust and reliable design must account for system uncertainties and exogenous disturbances, which may arise from phenomena such as modeling approximations, nonlinearities, degraded modes of operations, wind gusts, or structural impairments. Two widely recognized methods to mitigate the negative impact of uncertainties and disturbances are robust and adaptive control architectures. Specifically, robust control designs ensure stability by considering a worst-case scenario, which can potentially lead to suboptimal performance or steady-state errors depending on the nature of the uncertainty. In contrast, adaptive control designs offer the capability to learn and adjust to uncertainties, thereby reducing or eliminating the negative effects seen in robust control designs. In particular, adaptive designs adjust their control gains online, guided by desired performance, to counteract the impact of uncertainties.
The work presented here focuses on advancements in a class of adaptive control designs; hereinafter referred to as direct model reference adaptive control. This type of adaptive control design directly adjusts the control gains in real-time to allow the states of an uncertain system to track that of a known (user-defined) reference model, of which is constructed under the assumption that no system uncertainty or disturbance is present. This dissertation addresses four topics within the field of direct model reference adaptive control. The first topic is predictability in the system performance during the transient phase of operation through the concept of scalability. The second topic involves ensuring strict performance guarantees over a range of operating conditions via a gain-scheduled setting for a set-theoretic extension of model reference adaptive designs. The third topic focuses on enhancing system performance through the learning rate of the adaptive weight update law, which incorporates the time-derivative of the command profile. The fourth topic is time-varying command profile tracking through an optimal gradient term that adjusts the states of a reference model to approximately track a time-varying command profile of any form.
Scholar Commons Citation
Jaramillo, Jesse, "Advances in Model Reference Adaptive Control: Exploring Gain Scheduling, Scalability, and Time-Varying Command Following" (2024). USF Tampa Graduate Theses and Dissertations.
https://digitalcommons.usf.edu/etd/11132
