To more accurately describe the uncertainty of the DG power, this paper respectively establishes the output models of wind power DG and photovoltaic DG based on the Weighted Gauss Mixed Distribution (WGMD) and Beta distribution to solve the problem of reactive power optimization in the distribution network with Distributed Generations (DG). In the meantime, the Markov Chain Monte Carlo method combined with slice sampling algorithm is used for power flow calculation. The multi-objective reactive power optimization model which takes the minimum loss of the active power and minimum total voltage deviation of bus as objective function is established. And the improved Non- Dominated Sorting Genetic Algorithm (NSGA-II) is used to solve the optimization model. The feasibility and effectiveness of the proposed method is verified by the simulation of an improved IEEE33 node system. This work is supported by National Natural Science Foundation of China (No. 51867015 and No. 51767017).