Comparação de controladores pids robustos otimizados para conversores CC-CC

Abstract

This master thesis presents a comparison between two alternatives for the design of optimized robust proportional-integral-derivative (PID) controllers, with fixed control gains, computed offline, suitable for application in power converters with parametric uncertainties. Different from conventional designs, which obtain control gains for a nominal model of the plant and verify a posteriori the robustness against parametric uncertainties, the approaches used in this work guarantee robust performance in the design stage, for the plant with interval parameters. The first approach uses a linear programming problem, based on a target closed-loop polynomial and a linear objective function, to find the PID gains. The second approach uses a non-linear optimization algorithm, with an objective function based on specifications in the frequency domain, to find, through a meta-heuristic, the controller gains. Both approaches are guided by design criteria widely used in power converters, given by pole location, and crossover frequency and phase margin. The design procedures of the robust PID controllers are presented and illustrated by means of simulation and experimental results, allowing a comparison of the advantages and disadvantages of each technique.