Because static path planning (SPP) and rolling path planning (RPP) cannot solve the global optimal path, a global optimal path planning method (GOPP) considering the time-varying characteristics of road network weights was proposed. Vissim software was used to model and simulate a regional road network in Chongqing University Town, and the improved forward associated edge data structure was used to store the road network topology key elements and travel time simulation data, which were used as the path planning database. On this basis, the actual weights of cross period road sections were derived, and a GOPP method based on Dijkstra algorithm was proposed. Based on the path planning database, the classical Dijkstra algorithm was proved to have the global optimal solution ability compared with the intelligent heuristic algorithm. Finally, three planning paths in Matlab software were simulated by using the SPP, RPP and GOPP methods. The results show that the cumulative travel time of GOPP is 1 158.7 s, which is 212.7 s and 57.6 s less than those of SPP and RPP, respectively, verifying that GOPP is superior in shortening travel time. The proposed GOPP has certain theoretical significance for the development of intelligent vehicles in the future.