Controle adaptativo de corrente aplicado a conversores estáticos conectados à rede elétrica

Abstract

This thesis addresses the design of adaptive current controllers applied for voltagesource grid-connected converters with LCL-filter. Due to the grid equivalent impedance uncertainty and/or variation at the point of connection, the current controller of such converters may become unstable or have a poor performance with fixed-gain controllers because the grid impedance is part of the dynamic model. So, adaptive control techniques are proposed to ensure stability and performance of current controllers for grid-connected converter with LCL-filter even without knowledge of the grid equivalent impedance. Continuous time adaptive control, from the theoretical point of view, is a well established control technique. However, in this application, a discrete-time current controller is implemented. There are some differences between discrete-time and continuous-time adaptive controllers. When the controlled variable is the converter current, a model reference adaptive state feedback is proposed. The adaptation algorithm, responsible to update the gains in real-time, consists on a recursive least-squares (RLS) algorithm. When the controlled variable is the grid current, a model reference adaptive state feedback is also used. However, the RLS algorithm has some modifications to keep the controller robust under unmodeled dynamics. The RLS algorithm is used because it has better parametric convergence characteristics compared to other algorithms. The grid voltage disturbance is rejected making some small modifications on the controller structure. Simulation results are validated by experimental results in a 5,5kW three-phase grid-connected converter, which shows good convergence and steady state stability of the adaptive controllers. Finally, some comparisons with fixed gain current controllers are presented to show de advantages and disadvantages of each controller.