Abstract: In allusion to the uncertainty of renewable energy sources and loads, based on day-ahead planning and intra-day scheduling a robust optimization algorithm for islanded microgrid was designed. In this algorithm, aiming at the uncertainty of renewable energy sources and loads that might appear, the day-ahead planning searched the worst case scenario under the uncertain environment and utilizing column constraint generation algorithm the problem was decomposed into primal problem and subproblem, then by means of interactive iteration the optimal economic dispatch scheme under the scenario was solved; in the intra-day scheduling phase, utilizing current realtime measured information of renewable energy sources and loads and based on the dispatching results of day-ahead planning the realtime secondary optimization of the adjustment cost was performed, and the segmented punishment for wind power curtailment and PV power curtailment was implemented, thus renewable energy sources was furthest accommodated. Results of calculation example show that the designed algorithm can make the microgrid kept satisfied robustness and economy in the uncertain input and output environment brought by forecast error, so the effectiveness of the designed algorithm is verified.