Controle amostrado ótimo de sistemas lineares com saltos markovianos através de realimentação de estados

Author(s)

Gabriela Werner Gabriel

• Document type: Doctoral Thesis
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação
• Defense date: 2016-07-14

Examining board members

José Cláudio Geromel; Marcelo Dutra Fragoso; Marcos Garcia Todorov; Karl Heinz Kienitz; Roberto Kawakami Harrop Galvão; João Bosco Ribeiro do Val

• Advisor: José Cláudio Geromel

Abstract

This work is entirely devoted to develop an optimal sampled-data control law applied to Markov jump linear systems, whose main usage is Networked Control Systems (NCS). In this context, two network characteristics are simultaneously considered: the bandwidth limitation addressed by the existence of sampled-data signals in the system, and the packet dropouts modeled by a continuous-time Markov chain. In order to accomplish this goal, the general adopted approach is broken in four steps: stability analysis and norm evaluation based on the H2 norm; stability analysis and norm evaluation in the Hoo context; the optimal sampled-data control design that minimizes a J2 performance index based on the H2 norm, which can be expressed in a convex formulation based on LMIs; the optimal sampled-data control design that minimizes a certain Joo performance index based on the Hoo norm, which also admits a convex formulation based on LMIs, even though a deeper mathematical analysis is required. Each step has the same structure described in the sequel. First, the theoretical results are mathematically developed and proved. Second, some particular cases are derived from these theoretical results. Third, a convergent algorithm is proposed to solve each of the mentioned cases. The convergence of the algorithms are also proved. Finally, a numerical example illustrates the main developments in each step. The theory developed here is new and there is no similar result in the current literature. For a practical view of the outcomes, three practical examples are borrowed and adapted from available works: two of them are physical systems controlled through an NCS, where one is originally stable and the other unstable, and the third one is an economical system whose policy is applied in a discrete-time basis (AU)

• FAPESP's process: 12/23634-2 - Markov Jump Systems Control throw Communication Networks.
• Grantee: Gabriela Werner Gabriel
• Support Opportunities: Scholarships in Brazil - Doctorate