Defesa de Doutorado – Carla de Souza – 17/9/2021
Defesa de Tese de Doutorado | |
Aluna | Carla de Souza |
Orientador
Coorientador |
Prof. Eugênio de Bona Castelan Neto, Dr. – DAS/UFSC
Prof. Valter Júnior de Souza Leite, Dr. – CEFET/MG |
Data
|
17/9/2021 9h (sexta-feira)
Videoconferência (https://meet.google.com/uoa-mzut-xeh) |
Banca |
Prof. Eugênio de Bona Castelan Neto, Dr. – DAS/UFSC (presidente);
Prof. Alexandre Seuret, Dr. – MAC/LAAS/CNRS; Prof. Jéferson Vieira Flores, Dr. – DEEL/UFRGS; Prof. Daniel Ferreira Coutinho, Dr. – DAS/UFSC; Prof. Pedro Luis Dias Peres, Dr. – DSE/UNICAMP. |
Título | Control of Classes of Linear Discrete-Time Parameter-Varying Systems under Input Constraints |
Abstract: This work proposes control laws for the regional stabilization of two classes of discrete-time linear parameter-varying (LPV) systems under input constraints. The first one consists of systems subject to time-varying delay in the states, saturating actuators and energy bounded disturbances. For such a class, we establish convex conditions to design static state-feedback controllers as well as dynamic output-feedback controllers ensuring the regional input-to-state stability (ISS) of the control loop. The proposed conditions take into account the variation of the delay between two consecutive instants, which leads to better estimates of the region of attraction supporting higher levels of disturbances. The approach is based on the rewritten of the time-delay system with input saturation in terms of a switched augmented delay-free system with a dead-zone non-linearity. The second class comprises the systems with saturating actuators inserted in communication networks with limited bandwidth. In this case, we formulate convex conditions to synthesize event-triggering state-feedback controllers as well as event-triggering dynamic output feedback controllers ensuring the regional asymptotic stability of the closed-loop system while indirectly reducing the number of data transmissions over the communication channels. The proposed event-triggering policies indicate, for instance, whether the states or the output should be transmitted over the network or not. The use of the Lyapunov theory in all cases leads to conditions in the form of linear matrix inequalities (LMIs), which can be efficient solved by computational packages. Some numerical examples are provided to testify the validity and the effectiveness of our approach and to make comparisons with similar ones in the literature. |