Microgrid energy management optimization using model predictive control: a case study in China

In this paper, an energy management strategy using model predictive control (MPC) is applied to a Microgrid (MG) case study. Objective is to minimize the daily generation cost and emission of the MG while considering hard and soft technical constraints in real operations. The investigated MW-level MG is integrated in the industrial park of Goldwind Sci&Tech Co. Ltd. in Beijing, China and is composed of non-dispatchable power generation units (Wind & Photovoltaic), dispatchable generation units (Micro-turbines & backup Diesel generators), energy storages (Battery systems & Super-capacitor) and non-controllable local load. Normally, this micro-grid operates in gird-connected mode and external public distribution grid serves as auxiliary in adjusting the local power balance. With consideration of the time-of-use (TOU) electricity tariff in Beijing, a good planning and operation strategy of the dispatchable units as well as the energy storages can contribute to improving the global energy efficiency. As a result, a two-step optimization procedure is proposed for the microgrid: 1. a month-ahead planning optimization aims at deciding the maximal capacity threshold of power exchanges with the grid, which needs to be pre-purchased from the grid utility company; 2. a very short-time model predictive control, which is based on day-ahead prediction of the RnE generation and local load curves, aims at achieving the daily optimal control operations of the MG. Optimization models of both the two-step procedures is formulated in mixed-integer-linear-programming (MILP) scheme and the proposed energy optimization strategy has been validated by simulation results with support of actual data.