Microgrid Management with PV Power Prediction via Stochastic Distributed Optimization

Abstract

This paper is concerned with a stochastic distributed model predictive control (MPC) technique for power management of a PV-installed microgrid. The PV power supply has large uncertainty because it depends on weather conditions. To keep stable power supply to the microgrid, both accurate prediction of PV power supplies and efficient energy management based on the prediction are essential. We propose a distributed MPC method for microgrid control by combining the ADMM-based distributed optimization and the randomized algorithm approach under the situation that a stochastic prediction model for the PV power prediction is available. The proposed method enables us efficient energy management in a distributed way as well as the probabilistic guarantee of the line and battery capacity constraints. We demonstrate the effectiveness of the proposed method by a numerical simulation.