Síntese de controle robusto H [infinito] de ordem reduzida para sistemas lineares contínuos usando algoritmo genético e otimização por exame de partículas

Abstract

This dissertation is concerned with the problem of designing reduced order robust H∞ dynamic output feedback controllers, for uncertain continuous-time linear systems. The uncertain time-invariant parameters belong to a polytopic domain and may affect all the system matrices. The search for a reduced-order controller is converted into a problem of computing a static output feedback control design for an augmented system. To solve the problem, a two-stage linear matrix inequality (LMI) procedure is reviewed. It is analyzed, in this work, the effects of adding either a Genetic Algorithm (GA) or a Particle Swarm Optimization (PSO) metaheuristic procedures to the two-stage LMI procedure. The algorithms are calibrated using systems from a database known as COMPlib, and also randomly generated precisely known stabilizable systems, and the results are compared with a state-of-art synthesis technique, named HIFOO. As numeric experiments, randomly generated uncertain stabilizable systems and an uncertain system available in literature are used. The results provided by these techniques showed that they were capable to minimize H∞ norm and the results were close or better when compared to other techniques available in the literature.