Aiming at the problem of the waste of distributed generation resources caused by disorderly distribution of large-scale rooftop PV power stations connected to the rural distribution network, a multi- objective optimization configuration mode of large-scale roof PV power stations accessing to rural distribution network with comprehensive consideration of restrictions of installation area in rural region and operation constraints of distribution network, aiming to minimize the investment cost of rooftop PV power stations and minimize the loss of system network, is established in this paper. A kind of improved Elitist Non-dominated Sorting Genetic Algorithm (NSGA-Ⅱ) based on high-dimensional solutions for the optimization configuration model is proposed and the best configuration proposal is selected from Pareto solution set by applying fuzzy nearness. The proposed configuration method can improve the reliability of power supply and the economy of investment by using the optimization of rooftop PV power stations through simulation and analysis of distribution grid example of IEEE 33 node.