The optimization of feeder bus routes was studied to realize the effective connection between bus and subway and improve the transfer efficiency of the bus system. A two-layer programming model was established, considering the origin-destination (OD) demand of passenger flow and the transfer characteristics between multiple subway stations and bus stations. The objective function of the upper layer model aimed to minimize the sum of bus operating cost and passenger travel cost, and the integrity of the route and the rationality of the route were considered as two constraints. The lower layer model was a passenger flow allocation model, which constructed constraint conditions based on the line capacity and the station transfer. An elite retention strategy was introduced, and a model-solving algorithm was designed by combining the neighborhood search algorithm with the genetic algorithm. The case analysis indicated that the minimum error of the designed algorithm was 1.6%, and the algorithm efficiency improved significantly. After the optimization of the bus route network, the bus passenger capacity increased by 29%, and the per capita travel cost decreased by 13%. Experimental results show that the proposed model based on the principle of system optimization optimizes the bus stops around multiple subway stations. The optimization scheme has obvious advantages in improving the load factor, reducing the per capita travel cost and improving the transfer efficiency of the bus system.