Optimal Operation of an Energy Management System Using Model Predictive Control and Gaussian Process Time-Series Modeling

Abstract

This paper describes an optimal operation scheme for energy management systems using Gaussian process forecasting and model predictive control (MPC) in the context of grid-connected microgrids with local generation, loads, and storage. The main objective of the control is to minimize the cost of energy taken from the grid. The microgrid consists of a photovoltaic (PV) panel and a battery energy storage system, which are connected to a power grid and a local load via a dc bus. At each sampling time, the predictions for PV output power and load demand power are calculated, and an MPC algorithm is executed based on these predictions and a physical battery model to decide the set point of the battery. Simulations of two case studies, namely, a labscale microgrid and a commercial microgrid, are presented. We compare the performance of MPC with various horizon lengths to a rule-based control strategy to demonstrate a cost reduction of more than 2%.