For the problems of randomness, intermittence and correlation of distributed power sources such as wind and solar power generation, an improved semi-invariant probabilistic power flow calculation method is proposed to deal with the complex tail correlation between wind and solar fields and improve the calculation efficiency. Firstly, the probability model of random variables, the wind-solar output model and the joint distribution model of wind-solar are established. Secondly, considering the output correlation between wind and solar fields, an improved semi-invariant probabilistic power flow calculation method based on Rosenblatt transform is proposed, which greatly improves the efficiency of power flow calculation. Finally, the equivalent IEEE34 node system connected to the wind farm and photovoltaic field is simulated and tested, and compared with the Monte Carlo probabilistic power flow calculation method based on scenario reduction, which verifies the effectiveness and accuracy of the proposed method in the analysis of power flow characteristics of distribution network.