Heterogeneous fleet vehicle scheduling problems for dynamic pickup and delivery problem with time windows in shared logistics platform: formulation, instances and algorithms

his paper introduces a dynamic pickup and delivery problem with time windows, heterogeneous fleet vehicles, no depot, and dynamic priority (DPDP-TWHVNDDP) in the shared logistics platform. The distinguishing feature of DPDP-TWHVNDDP is that the priority and number of requests and the number and types of vehicles available are dynamic, and that all the requests in a scheduling are not necessarily served to maximize the total profit according to their priorities. A mathematical formulation of maximizing profit dynamic scheduling adapting to dynamic priority is established, and a hybrid scatter search parallel algorithm combined with the genetic algorithm, variable neighborhood search algorithm, and tabu search algorithm is proposed to solve the problem with high quality in a relatively short time. Comparison results of our algorithm against the best-known solutions on the benchmark instances available in the literatures show that our algorithm can improve over 35% (52 out of 146) of the total distance and over 30% (44 out of 146) of the vehicle used. A set of DPDP-TWHVNDDP instances based on extending Li and Lim's benchmark instances are introduced. Experiments on benchmarks and actual datasets show that the proposed hybrid heuristic is effective and efficient especially in large scale datasets.