Abstract: Uncertainty of renewable energy generation power and load usage is an important factor affecting the optimal dispatch of microgrid energy. On the premise that these uncertainties are expressed by intervals, researchers proposed a two-stage robust optimization method based on the worst scenario. Considering the probability of these uncertainties prediction information will be at the worst value in each period is very small, and in most cases these will be near the expected value, a two-stage robust optimal dispatching model for microgrid based on the expected scenario was proposed to improve the economic efficiency of the system operation: it aims to determine the first stage（pre-scheduling stage） decision-making results for the optimization objective of system economy in the expected scenario, and to ensure that the feasible second stage（re-scheduling stage） decision-making results can be obtained even in the worst scenario based on the determined decision variables in the first stage, so as to achieve the "expected optimal, worst feasible" goal of system optimization. Based on the two-stage zero-sum game theory, an improved column and constraint generation（C＆CG） algorithm was proposed. Experiments show that compared with the classical two-stage robust optimal scheduling model based on worst-case scenario, the proposed model based on expected scenario can not only run steadily in uncertain intervals, but also has lower operating cost.