In order to reduce the waiting time caused by the flight delay, an route optimization model for the customized airport bus was established considering the uncertainty of the road network and the trip demand of the customization and differentiated. Then the maximum operating income, the minimum cost of vehicle and the minimum penalty cost for early arrival of vehicles were proposed as objective function, which was solved by differential evolution algorithm. To avoid algorithm premature, an improved adaptive operation method was introduced to enhance the ability of the global optimization. The calculation examples show that the waiting time of passengers can be reduced from 26.61% to 46.68%, when the dynamic path optimization model is used considering the flight delay and the booking of road network. The model is considered as reliable to optimize the route of airport bus.