Defesa de Exame de Qualificação – Marco Aurélio Schmitz de Aguiar – 12/12/2017
12/12/2017 17:10
| Defesa de Exame de Qualificação | |
| Aluno | Marco Aurélio Schmitz de Aguiar |
| Orientador
Coorientador |
Prof. Eduardo Camponogara, Dr. – DAS/UFSC
Prof. Morten Hovd, Dr. – NTNU/Noruega |
| Data
Local |
12/12/2017 10h30 (terça-feira)
Sala PPGEAS I (piso superior) |
| Prof. Rodrigo Castelan Carlson, Dr. – DAS/UFSC (presidente)
Prof. Jorge Otávio Trierweiler, Dr. – ENQ/UFRGS Prof. Hector Bessa Silveira, Dr. – DAS/UFSC |
|
| Título
|
Distributed Otpimal Control of Nonlinear Dynamic Networks |
| Abstract: Optimal control is a field that looks into finding the best controls to apply to a system with respect to some given criteria. In the last century, a lot of effort has been put into creating a strong theoretical background. Lately, the advances in computational power and specific software made it possible to solve optimal control problems for a range of systems and applications. In this work, we are interested in making scientific and technological contributions in the specific area of distributed optimal control applied to distributed systems. Distributed systems are dynamic systems that are composed by many coupled subsystems. These systems occur naturally in various fields: consumers and generators in a power supply network; reactors, valves, and tanks in chemical plants; interconnected autonomous vehicles; etc. Developing centralized approaches, where a single regulator controls the whole system, might be too costly for large scale, and at the same time it creates a single point of failure. On the other hand, a decentralized approach where each subsystem has its own controller that is independent of the other subsystems can lead to poor performance or even unstabilize certain systems. Therefore, in this work a cooperative distributed controller is proposed where each subsystem has its own controller but also accounts the interactions of neighboring subsystems. To achieve this, a framework for modeling networked systems is proposed, the framework renders a decoupled cost and coupled constraints model, where the couplings are simply the equations that connect inputs and outputs of neighboring subsystems. This framework facilitates the modeling and creates a structure that is highly sparse. To exploit the structure created by this framework, two algorithms are proposed for cooperative optimal control of distributed systems. Later a modification to these algorithms is proposed to achieve full parallelization of the computation of the optimal controls. | |
