Controle MPC com amortecimento ativo aplicado a um VSC com filtro LCL conectado à rede

Abstract

The installation of distributed generation systems, such as photovoltaic energy and wind energy, connected to an electrical grid is in full growth and, for its connection, power converters are necessary together with control methods. Currently, the use of predictive control is in full development, mainly for applications in power electronics. There are several types of predictive control, however the model predictive control (MPC) is addressed in this dissertation due to its fast dynamic response, ease of including restrictions and nonlinearities, makes the system fully decoupled, considers multivariable cases and presents easy implementation. Thus, the present work presents deal with the MPC applied to a grid-connected converter with LCL filter. The MPC predicts the future behavior of the controlled variables and with this information chooses the best switching state according to a cost function optimization. However, before applying the MPC method it is necessary to develop the mathematical model of the system, where the state space method was used for later discretization. The equations that describe the model were first written in the abc natural coordinates and later transformed to the αβ stationary coordinates using the Clarke transform. In order to avoid the intrinsic resonance of the LCL filter, a design of an active damping strategy based on virtual resistor is detailed, as well as the inclusion of this restriction in the predictive control cost function. In addition, disturbances were inserted into electrical grid and techniques in order to mitigate them were proposed and applied. Finally, simulation results are used to validate the proposed method and to show its effectiveness.