Model Predictive Control of Energy Management System for Economic Dispatch with Application to MHS Microgrid in Normal Operation

Power generation in microgrid are conventionally decentralized making the grid management increasingly complex. Energy management system (EMS) is very important for power system planning and operation of microgrid. To make sure that the generation meets challenges of load profile requires more robust and effective EMS of microgrid. Specifically, the planning and operation of power supply needs to incorporate the load forecast for the optimal dispatch strategy. This paper proposes Model Predictive Control (MPC) of EMS to dispatch generation power in microgrid for minimum total operating cost. The load forecast is obtained by recurrent neural network for one day ahead. The efficiency and accuracy of load forecasting model affects the efficiency of dispatch algorithm of microgrid. The proposed algorithm is applied to a case study of Mae Hong Son (MHS) microgrid including battery energy storage system and power exchange with utility grid to satisfy load demand for both actual load and forecasted load profiles. We apply MPC to dispatch energy in normal operation and compare the total operating cost (TOC) with that of the previous EMS. It is found that the proposed MPC can significantly reduce TOC ranging from 25% to 45%.