Abstract: In view of the shortcomings that the influence of the diversification of load characteristics on the photovoltaic absorption capacity of the power system cannot be fully considered in the process of evaluating the photovoltaic absorption capacity of the distribution network,a method of photovoltaic absorption simulation and probability evaluation considering the actual situation in load operation is proposed. Firstly,the support vector machine-random forest algorithm is used to clean the load data,and then the improved spectral clustering algorithm is used to cluster the different types of load characteristics. Furtherly,the type curves of different loads are selected as the load data at nodes based on the scenario analysis method. Finally,based on the Monte Carlo simulation method and the uniform sampling method,the stochastic simulation of the photovoltaic absorption scheme and the approximate evaluation of the absorption capacity are carried out. Based on the example analysis of IEEE33 node system,the sampling results considering the load characteristics are compared,which effectively verifies the applicability of the proposed method.