Full text
| |
| Author(s): |
Rafael Fontes Souto
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação |
| Defense date: | 2010-03-31 |
| Examining board members: |
João Bosco Ribeiro do Val;
Marcelo Dutra Fragoso;
Pedro Luis Dias Peres
|
| Advisor: | João Bosco Ribeiro do Val |
| Abstract | |
This dissertation presents a framework for continuous-time stochastic systems in which the control variations increase the state uncertainty. This type of system can be applied in several areas of science and engineering, due to its hability of modelling complex stochastic systems, for which the dynamics are not completely known. Controlled Itô diffusion processes are used in order to describe the state path, and the optimization was achieved by the dynamic programming method, so it was necessary to solve the Hamilton-Jacobi-Bellman equation. In addition, tools from nonsmooth analysis indicated the existence of a region in the state space in which the optimal control action is characterized by no variation, no matter the previous control were. Intuitively, this result is expected from the cautionary nature of controlling underdetermined systems. Finally, it was analytically studied the particular case of a system with quadratic running costs. This study revealed that the technique developed allows the computation of the optimal solution in a simple and effective way for asymptotic behavior of the system. This feature of the solution comes in hand to obtain the complete solution of the problem by means of numerical approximations (AU) | |
| FAPESP's process: | 07/06293-9 - Output Feedback Control of Markov Jump Linear Systems |
| Grantee: | Rafael Fontes Souto |
| Support Opportunities: | Scholarships in Brazil - Master |