A Model Predictive Control (MPC) strategy based on the Evolutionary Algorithms (EA) is proposed for the optimal dispatch of renewable generation units and demand response in a grid-tied hybrid system. The generating system is based on the experimental setup installed in a Distributed Energy Resources Laboratory (LabDER), which includes an AC micro-grid with small scale PV/Wind/Biomass systems. Energy storage is by lead-acid batteries and an H2 system (electrolyzer, H2 cylinders and Fuel Cell). The energy demand is residential in nature, consisting of a base load plus others that can be disconnected or moved to other times of the day within a demand response program. Based on the experimental data from each of the LabDER renewable generation and storage systems, a micro-grid operating model was developed in MATLAB(C) to simulate energy flows and their interaction with the grid. The proposed optimization algorithm seeks the minimum hourly cost of the energy consumed by the demand and the maximum use of renewable resources, using the minimum computational resources. The simulation results of the experimental micro-grid are given with seasonal data and the benefits of using the algorithm are pointed out.