UC Santa Barbara

The focus of this thesis is on developing control strategies for large-scale systems. We look at two distinct problem areas: the design of coordination algorithms for multi-agent systems and the optimization of industrial refrigeration systems.In the first part of this thesis, we focus on multi-agent systems, where various decision- makers interact with each other, each with their own local objectives. To understand and design control strategies within these systems, we employ a game-theoretic viewpoint. Under this framework, the utility functions and strategic interactions are explicitly modeled to understand emergent behavior. Optimal incentive mechanisms can then be derived to align joint outcomes with the collective objective. The subsequent chapters explore different scenarios within this framework, such as collective transient behavior, coordination in limited information settings, and run-time analysis. Overall, we aim to classify multi-agent behavior and optimize outcomes under varying conditions. In the second part of this thesis, we focus on designing control strategies for industrial refrigeration settings. Industrial refrigeration plays a significant role in various sectors and represents a major portion of total global energy usage. Despite this, there are significant control opportunities in increasing the efficiency of such systems by carefully modulating system variables, such as pressure and temperature, to operate close to optimal thermodynamic conditions. We explore the control opportunities of compressor sequencing and scheduling as a viable option to significantly reduce energy usage, by utilizing techniques from inventory control and scheduling theory. By leveraging optimization ix methodologies, the manuscript seeks to enhance the efficiency of refrigeration systems, thereby reducing energy usage and environmental impact.