In order to configure distributed generation (DG) more reasonably, an optimal DG configuration model for distribution network considering voltage weighting factor is proposed. The objective functions of the model are the total voltage deviation and the total cost including DG investment cost, operation and maintenance cost, network loss cost and power purchase cost. By introducing the voltage weighting factors into the total voltage deviation objective function of the model, the influence of DG configuration on the voltage of each node and the evaluation of the voltage quality requirements of users are reflected. The voltage amplitude of the node with larger voltage deviation or with higher voltage score is easier to approach its expectation. DG and load status are obtained by Latin hypercube sampling. NSGA-II algorithm and grey target decision algorithm based on entropy weight are used to get the optimal solution. The simulation example shows the feasibility and effectiveness of the proposed method.