Considering the reactive power regulation and control ability of the distributed generator(DG)

the stochastic optimal reactive power flow(SORPF)model is proposed

in which the state variables are constrained by chance constraints. Nonparametric kernel density estimation is used to obtain the probability density function of random factors. The probabilistic power flow(PPF)calculation based on the stochastic response surface method is analyzed

which can be suitable for a variety of probabilistic models. The variables' correlation processing method based on Nataf transformation of PPF is also proposed. The SORPF model can be solved by referring to particle swarm optimization. Finally

the modified IEEE 33-bus distribution system and American PG&E 69-bus distribution system are utilized to test the correctness and effectiveness of SORPF. It follows from the test results that the proposed PPF method has higher accuracy when computing the cumulative distribution function. The test of different scenarios further verifies the model and algorithm proposed in this paper can adapt to different DG control strategies

find the probabilistic cross-border risk and have the ability to fix it.