As urban traffic becomes more and more complex, time and path cost directly determine the effect of path planning. The traditional A* algorithm cannot satisfy these needs any more. We thus propose a TWA* algorithm, which can improve its evaluation function on the basis of the original A* algorithm. Firstly, a time factor normalization model is established by time parameter to expand the probability of node selection and save time cost. Secondly, it is combined with the valuation function to reduce the distance cost. The A* algorithm and TWA* algorithm are respectively verified in the experiment using the GPS data of a region of Beijing. The results show that the TWA* algorithm improves the time and path cost  by about 6% and 5% respectively, achieving the goal of path optimization and providing reliable reference basis for efficient delivery of enterprise logistics.