Abstract: Aiming at the problem of the large number of unknown coefficients in the polynomial expansion is large and the algorithm easy falling into the dimension disaster when the traditional stochastic response surface method is applied to high dimensional input variable system,an improved probabilistic power flow calculation method is proposed. By analyzing the first-order sensitivity coefficient of input variables to the output responses of power flow calculation to delete the quadratic cross terms in the second-order SRSM polynomial,which can obtain the sparse expression form of the SRSM multinomial expansion,and the Nataf transform technique is used to convert non-normal variables（such as wind speed）into normal variables as system input. The effectiveness of the proposed algorithm is verified by simulation on the IEEE-9,IEEE-118 standard node systems with wind power. The simulation results show that the proposed method can significantly reduce the number of deterministic power flow calculations compared with the traditional stochastic response surface method,It can effectively solve the problem that the traditional stochastic response surface method is difficult to be applied to high dimensional input variable systems.