Abstract: In order to solve the cost waste caused by the change of urban traffic road information or customer demand, a dynamic vehicle routing problem model and solution algorithm with time window and capacity constraints are proposed. Firstly, a dynamic vehicle routing problem with time windows and capacity constraints is established to minimize the total vehicle cost, and DVRP is decomposed into a set of VRP problems by DVRP solver to solve the dynamic problem. Secondly, the coordinate firefly algorithm is proposed to map the discrete solution of the firefly algorithm to the continuous domain to be suitable for the model. The local search is used to improve the algorithm, including initial population, enhanced path, removing nodes and exchanging nodes. Finally, combined with data sets and examples, Matlab is used to analyze the performance of the algorithm. The results show that the proposed algorithm is significantly better than the classical DVRP algorithm in both solution speed and solution quality, and the practical significance of the algorithm and model is verified by an actual case.