In order to improve the capability of renewable energy accommodation, an economic dispatch model for microgrid based on time-scale decomposition of energy storage is proposed. This model includes two stages: day-ahead scheduling stage and real-time dispatch stage. Adopting new type compressed air energy storage in day-ahead scheduling stage, an economic dispatch optimization model with the objective of minimum comprehensive cost is established, which takes account of economic cost, environmental benefits, price difference between peak and valley, and life cycle cost of energy storage. In real-time dispatch stage, the operational schedule is corrected timely according to the forecast errors of renewable energy generation, while the battery capacity is allocated to meet a certain confidence level. According to the case study of a microgrid demonstration project, the results verify that the proposed energy storage is applicable for peak load shifting, and it can effectively smooth the fluctuations of renewable energy. Finally, compared with different energy storage schemes, the results show that the proposed energy storage has enormous advantages in feasibility and cost.