Abstract: Reasonable characterization of source load fluctuation is the key to economic operation of DC microgrid. Existing studies often regard source load as an independent distribution value, without considering the continuity of source load in adjacent periods, which makes the characterization result deviation from the reality and affects the consumption of renewable energy. Therefore, based on the time dependent characteristics of the source power, this paper proposes a novel day-ahead to intra-day robust optimization scheduling strategy for DC microgrids.Firstly, through the historical data of the source load, it is proved that the power of the source load has correlation characteristics in the adjacent time periods. The Gaussian process regression algorithm is used to reasonably depict the time dependent characteristics of the source load, construct the uncertainty set of the source load, and avoid the unreasonable scene affecting the operation decision. Secondly, due to a certain deviation between the severe scenario of the robust optimization and the real-time power of the source load, the supply and demand of the source load do not match, and tripping power failure may occur. In order to solve this problem, this paper establishes the intra-day robust optimization scheduling, and uses the energy storage system and the bidirectional grid-connected inverter as the flexible scheduling means to realize the revision of the day-ahead scheduling decision and ensure the stable operation of the DC microgrid. Finally, the effectiveness and superiority of the proposed strategy are verified by comparison of numerical examples.

    Key words: source load fluctuation; time dependence; DC microgrid; day-ahead to intra-day robust optimization scheduling