This thesis presents a model predictive control (MPC) approach to economic scheduling for a building microgrid at California State University, Long Beach. First, components of the microgrid relevant to operational costs are modeled. Next, a peak demand cost model to extend MPC-based microgrid energy scheduling is proposed. The corresponding objective function is then formulated as a mixed-integer linear programming (MILP) problem. The MPC framework is implemented onto MILP optimization to construct MPC-MILP, which is formulated to compensate for uncertainties in day-ahead demand, photovoltaic (PV) power forecasts and system modeling. Next, the forecast modeling for demand and PV power to improve the accuracy of MPC-MILP is provided. The simulation results show that the MPC-MILP optimization approach provides superior cost minimization over other strategies such as MILP, which controls the microgrid subject to only one calculation using day-ahead forecasts. |