Abstract: Based on the survey of harbor-shaped bus stops in the main roadways in Nanchang City ,this paper gen-erates a large quantity of traffic flow data from Vissim simulation .A neural network is built from the six parameters of the bus stop length ,number of waiting passengers ,number of lanes ,road vehicle average speed ,number of vehicles stopped in the stop ,and the import and export of the number average delay time ,etc .The ratio of section average vehicle delay to link capacity is used as the capacity reduction .The neural network is implemented in Matlab platform to obtain variable factors and output influence weights and bias values of connection strength of W and B ,which provides the quan-titative influence coefficient for the bus stops on the road capacity .