Combining Heuristic and Exact Methods to Solve the Vehicle Routing Problem with Pickups, Deliveries and Time Windows

摘要

The vehicle routing problem with pickups, deliveries and time windows (PDPTW) is an important member in the class of vehicle routing problems. In this paper a general heuristic to construct an initial feasible solution is proposed and compared with other construction methods. New route reconstruction heuristics are then shown to improve on this. These reconstruction heuristics look to reorder individual routes and recombine multiple routes to decrease the number of vehicles used in the solution. A tabu search scheme where the attribute to be recorded has been specifically adapted to the PDPTW is proposed. A new method based on branch and bound optimisation attempts to optimise the final ordering of requests in routes to further improve the solutions. Results are analysed for a standard set of benchmark instances and are shown to be competitive with the state of the art.