An algorithm of dynamic vehicle scheduling problem based on approximate dynamic programming

Vehicle scheduling in service industry of logistics distribution was presenting features including the tasks tended to be of large scale, vehicles were multi-type and had multiple attributes as well as high demands for real-time scheduling. To solve these problems, this paper proposed a dynamic vehicle scheduling algorithm based on the approximate dynamic programming. An approximate value function was obtained through training of some samples, and according to mission requirements, vehicle state and conditions, and quick scheduling decisions could be made with the value function. The simulation test has proved the correctness and effectiveness of the algorithm.