Abstract: Based on the analysis of the transportation network design problem,the discrete transportation network design problem was discussed.The bi-level programming model was used to describe the problem,and the upper model considered the point of view of the transportation planner,in order to design the optimal network structure and achieve system optimal based on the restriction of budget,the lower model considered the user of the network and in order to attain the user optimal.The simulated annealing algorithm(SA) was used to solve the upper model and the gradient projection algorithm(GP) was used based on the path search to solve the lower model.The simulated annealing algorithm designed obtains satisfying result and has strongly robustness,which can solve large scale traffic network design problem.GP algorithm meets the essential characters of Wardrop equilibrium and gains more efficient by steady adding path.Application denoted the optimization method was rational,the result can improve the transportation status evidently,and the convergence speed of the algorithm is extraordinary rapid,so it provided scientific guidance for the decision-making of the transportation network planning.