Abstract: An optimal scheduling method of a building microgrid with the virtual storage system（VSS） was proposed based on the model predictive control approach. Firstly, a building virtual storage system was modelled and scheduled as a flexible and controllable unit under the model predictive scheduling framework. Then a model predictive control method of the building microgrid was developed. By using the repeated rolling optimization scheduling with finite horizon instead of the one-off optimization scheduling with the whole period, the problem that the prediction accuracy of renewable energy sources decreases with the increasing time scale was resolved effectively. Finally, two different type of building Microgrid cases under the summer refrigeration scenario were carried out to demonstrate the effectiveness of the proposed method. Numerical studies demonstrate that the proposed optimal scheduling method can reduce the operating cost of the building microgrid by making full use of the thermal mass of the building. At the same time, the method has better control performance and stronger robustness with the uncertainties from the forecasting data being considered. It can effectively solve the errors between the schedules of the building microgrid at the day-ahead stage and the actual operation stage.