The multiple-depot vehicle routing problem with vehicle share and the genetic algorithm

In the MDVRP system with vehicle share, the vehicles can come to any depot. And because the fuel cost becomes higher and higher, the different cost between full-load vehicle and unload vehicle can not be ignored anymore. So the new MDVRP model must meet the new situation, and the objective function must include the fuel cost of different path, and the vehicle's path constraints do not have to come back to the original depot. Then because the model is still a NP problem, an improved chromosome representation is proposed on the basis of path in this paper. A typical result and the analysis of experiment indicate the validity of the method to solve the MDVRP.