A day -ahead scheduling method considering the flexible supply and demand balance of distribution network after large-scale distributed photovoltaic (PV) access is proposed. Firstly, the hybrid Gaussian model is used to establish the probability distribution model of PV and load prediction error, and the robust interval of distribution network scheduling flexibility demand is obtained by generating prediction error scenarios based on Monte Carlo simulation method. Then, a distribution network day-ahead scheduling model aiming at minimizing the daily operating cost and the robust balance penalty cost of flexible supply and demand is established, and the improved Benders decomposition method is used to solve the problem. Finally, the IEEE33-node system is taken as an example to verify the effectiveness of the proposed method.