Investigation of model predictive control for converter-based stand-alone DC distribution networks fed by PV units

Abstract

Summary This paper applies model predictive control to converters of a stand-alone direct current (DC) distribution network. The proposed network is similar to a real stand-alone network and includes essential components from generation to consumption. The network uses photovoltaic (PV) units as sources of electrical energy. Each PV unit is connected to the DC network via a DC-DC boost converter. Both alternating current and DC load types are included and fed through 3-phase inverters and a DC-DC buck converter, respectively. As the network operates in stand-alone mode, an energy storage system is considered. The energy storage system is consisted of a battery bank and a bidirectional DC-DC converter to regulate and control both the network's voltage level and the operation of PV units. Model predictive control-integrated diagrams have been designed separately for each converter. It is expected to have continuous power flow from PV units to loads during daytime and nighttime and a regulated voltage level for the network controlled by energy storage system. Simulations done by PSCAD/MATLAB interfacing are used to demonstrate how power flows through the network and is consumed by loads or stored in batteries. The feasibility of model predictive control to control converters for this application is concluded through comparison of results with classic controllers' performance under equal conditions.