Abstract: In the planning and operation of current power system,a number of uncertainties have been brought into the power system because of electric components’ operating state,data collection and multipoint access of renewable energy,etc.Of these uncertainties,the most prominent is the uncertain power flow,which will fluctuate when stochastic fluctuation is generated by the nodes’ injected power.To solve this problem,an improved multi-point estimate method of solving uncertain power flow based on Nataf transformation and Gauss-Hermite integration is proposed.The basic idea is to change the uncertain problem of power flow into a response function with multiple random variables.Firstly,the multi-estimate points in the standard normal space and the relevant weight can be obtained according to Gauss-Hermite integration.Secondly,the estimate points in the original variable space can be obtained by Nataf transformation.With the estimate points in the original variable space and the relevant weight,the node voltage amplitude and phase angle,the branch active and reactive power can be easily calculated statistically.Simulations show that the results obtained by the improved multi-point estimate method have a higher precision compared with the traditional point estimate method.In addition,the precision of the calculation results will be improved if the number of selected estimate points is increased.