Abstract: This study presents a comparative study of a Generalized Predictive Controller (GPC) with a classical Proportional-Integral-Derivative (PID) controller and the Model Predictive Controller (MPC). All these strategies are compared on their output response, steady state error, controller effort and disturbances rejection. These controllers were designed and evaluated to be implemented and work on a micro grid DC system located at the Universidad Militar Nueva Granada Campus in Cajica, Colombia. Disturbances in this study were predicted by Nonlinear Autoregressive with Exogenous Excitation Artificial Neuronal Networks Model (NARX).