Moving window optimization with flexible temporal structure for operating grid-connected photovoltaic microgrids

Utilization of prediction data is vital for the efficiency of operating microgrids with renewable generations. In this paper, we modify a double-layer optimization approach for optimally operating photovoltaic (PV) dominated microgrids in grid-connected mode. The approach consists of an upper schedule layer and a lower dispatch layer. Based on a moving optimization window, the schedule layer periodically produces operating reference, which is traced by microgrid units in real-time dispatch layer. The moving window has a novel flexible temporal structure (FTS), which enables schedule layer to utilize prediction data at multiple time scales for generating power reference. As a byproduct, the optimality of operational cost and size of optimization problems are able to be controlled by adjusting the temporal structure. Effectiveness of FTS is validated by comparative numerical simulations on a 14-bus grid-connected microgrid, which includes PV units, energy storage systems (ESS), distributed generators (DG) and loads.