According to the characteristics of islanded microgrid with renewable energy and load uncertainty, a robust economic dispatch algorithm for islanded microgrid is proposed based on echelon deviation uncertainty set. The microgrid model is set up for its structural characteristics, and the uncertainty set is converted into the uncertainty set with echelon deviation. In the day-ahead scheduling stage, the outputs of each device of the microgrid are optimized to obtain the day-ahead scheduling scheme with the optimal solution for all scenarios feasible and performance of the expected set of scenarios optimal. In the intra-day scheduling stage, it maintains the power output of the energy storage system obtained in the day-ahead scheduling stage, and adjusts the day-ahead optimal solution of micro-gas turbine, curtailment of photovoltaic power and wind power, and shiftable load to improve the economic performance through a secondary optimization utilizing the real-time data of new energy and load . The results show that the proposed method achieves better control performance and improves the economy of microgrid dispatch. Finally, the effectiveness of the method is verified by a simulation case.