Abstract: Aiming at the typical day selection problem in microgrid capacity planning, this paper comprehensively considered the total characteristics, distribution characteristics, typicality and extremity of resources and load, and constructed a comprehensive evaluation index system of typical day selection, which was combined with mix integer multi-objective linear optimization to establish a typical day selection model. This model used a two-stage fuzzy plan to assign typical day weights, which reduced the error between typical days and origin data. Based on the typical day selection, a bi-level planning model for microgrid was established. The upper model aims to maximize the net profit of the microgrid operator and optimize the configuration of the wind-solar and storage capacity. The lower-level model considers the economic dispatch of the microgrid. The typical daily selection results based on the comprehensive evaluation index system can further improve the accuracy of the planning results and effectively reduce the planning and design time. Finally, the measured data of a certain area in western China was selected to verify the effectiveness of the proposed model and the solution.