Unbalanced load, asymmetric network parameters and access to renewable energy are being connected to distribution network, three-phase probability load flow is introduced. In this paper, a new three-phase probabilistic power flow calculation method is proposed, and the RBF neural network and the unscented transform method are used to solve the three-phase probabilistic load flow. Firstly, the Sigma matrix and the corresponding weights of the input variables are obtained by using mean and covariance matrix of input variables according to unscented transformation method. Secondly, the RBF neural network is used to solve the nonlinear equations of the power flow, and the mean of the output variables and the corresponding weights are obtained. The proposed method is suitable for the input variable with correlation. This ability improves the speed of algorithm due to not requiring calculating the Jacobian matrix and its inverse matrix. Finally, the unbalanced 25-bus system was tested in in the simulation calculation. The results show the effectiveness and practicability of the proposed algorithm.
