Optimal energy management via MPC considering photovoltaic power uncertainty

In this paper, we propose a method that uses model predictive control (MPC) to predict photovoltaic (PV) power generation, plan for the electricity demand in a building using the predicted value, and apply it online to correct the prediction error. First, we construct the regression model using a PV experimental unit and past data obtained from the Meteorological Agency. Next, we predict the PV power using grid point power (GPV) data of the next day. Second, the air conditioning or heating of the building is modeled to determine the electricity demand so that it increases the profits to the consumer and reduces the peak in time-varying electric cost. The error between the predicted and true value is considered via MPC. Finally, we show the advantages of the proposed method by performing simulations.