Sintonia automática baseada em otimização por enxame de partículas aplicada a controladores de corrente de conversores conectados à rede

Abstract

This work presents procedures of automatic tuning of state feedback controllers applied to grid-connected converters. The methodology employed here is comprised by the following steps: i) modeling the converter and the controller using state space description; ii) proposition of performance indices based on the tracking error and on constraints based on the closed-loop eigenvalue positions, taking into account parametric uncertainties and limits of the control signal; iii) employment of the particle swarm optimization algorithm in order to find the control gains by means of a search in large spaces, aiming at the minimization of the cost function and not violation of the design constraints; iv) validation of the control gains designed by means of simulations with increasing levels of fidelity, from simulation of simple linear models of the converters to simulations of the converter in hardware-in-the-loop. The design procedures are executed off-line, using the particle swarm optimization algorithm, which finds, in reasonable computational time, state feedback control gains that lead to results of good quality. The real time simulations show good tracking of sinusoidal references and good rejection of harmonics, providing grid currents complying with a standard pertinent to grid-connected converters. The experimental results in a prototype of industrial pattern prove the effectiveness of the proposal.