In order to improve the photovoltaic capacity of microgrid and tap the flexible adjustment potential of load side new balance resources, a differentiated aggregation method based on improved isomorphic polytopes is proposed. The operation constraints of flexible resources such as energy storage, temperature-controlled loads and so on are uniformly represented as linear polytopes. After rule-based grouping, the inner approximation aggregation is realized through the scaling and translation of the benchmark isomorphic polytopes; According to the characteristics of photovoltaic high noon power generation, the time-differentiated weights are introduced to enhance the regulatory capability of core time periods, and the day ahead optimization model of microgrid with photovoltaic consumption as the guidance and minimum operation cost as the goal is constructed. The simulation results show that the proposed method can accurately characterize the aggregation flexibility, significantly improve the PV consumption level, stabilize the output fluctuation, and the scheduling feasibility and calculation efficiency meet the engineering requirements, providing an efficient solution for the large-scale load side resource support of new energy consumption in the microgrid.