Research On Distribution Routing Of Replenishable Unmanned Distribution Vehicle Carried By Truck

With the increasing social demand for logistics distribution,the labor intensity of logistics practitioners continues to increase.Meanwhile,rising labor costs and the rapid development of autonomous driving technology make unmanned delivery a future trend.Currently,most of unmanned delivery vehicles are powered by batteries.Due to existing technical limitations and policy considerations,these unmanned delivery vehicles are temporarily not suitable for long-distance and large-capacity transportation.If traditional trucks are used to transport unmanned delivery vehicles to the last-mile delivery area,it can make up the shortage of unmanned delivery vehicles to a certain extent.Considering the characteristics of large amount of goods distributed in cities,limited number of unmanned delivery vehicles and scattered parking points,this thesis studies the problem of using a large vehicle(LV,truck)to carry some small unmanned delivery vehicles(SUAV)for combined distribution.That is,a truck is equipped with multiple SUAVs.The LV is responsible for transporting,picking up and releasing SUAVs.The SUAVs deliver the goods to customers,and can replenish goods at the LV and perform multi-trip delivery.The main research contents of this thesis are as follows:(1)Combined with the actual background,a basic model of distribution routing of replenishable unmanned distribution vehicle carried by truck without time windows is established.The model takes into account the situation that the SUAV is multi-bin format where one compartment can only store one item,the LV can carry,release and pick up the SUAV,the LV can replenish the SUAV with goods at the parking point,and the SUAV can perform multi-trip open distribution after the goods replenishment.With the objective of shortest total distribution path,a mathematical model with the constraints of coordinated distribution between LV and SUAVs is established.(2)A hybrid genetic algorithm with large neighborhood search is designed for the distribution routing of replenishable unmanned distribution vehicle carried by truck without time windows.Based on the genetic algorithm,a large neighborhood search algorithm is added to optimize the individuals.In the algorithm optimization process,the SUAV path is optimized first,and then the LV path is optimized on the basis of the SUAV path.On the modified Solomon data,the proposed hybrid genetic algorithm with large neighborhood search is compared with Gurobi solver,genetic algorithm,and large neighborhood search algorithm to prove the efficiency of the algorithm.The proposed algorithm uses the shortest time to obtain 8 sets of best solutions in 9 sets of experimental data.(3)For the distribution routing of replenishable unmanned distribution vehicle carried by truck with time windows,a model is formulated and an adaptive large neighborhood search algorithm is developed to solve the model.In the algorithm,an adaptive mechanism based on the betting mechanism and an insertion operator with evaluation of the number of insertable positions are innovatively designed.And a depth-first search strategy with pruning is adopted to generate a LV path based on the existing SUAV path.(4)Experiments and analyses are carried out for the distribution routing problems of replenishable unmanned distribution vehicle carried by truck with time windows.First,the Taguchi method is used to adjust the parameters of adaptive large neighborhood search algorithm.Then,the adaptive large neighborhood search algorithm is compared with the hybrid genetic algorithm with large neighborhood search,Gurobi solver and other heuristic algorithms by the experimental comparison analysis on the modified Solomon datasets.Results show that the proposed of hybrid genetic algorithm with large neighborhood search and adaptive large neighborhood search algorithm are effective in solving the problem.For small-scale problem instances,the difference between the results obtained by the two algorithms is within 0.5%,and as the data size increases,the remains around 2%.In addition,statistical analysis shows the significance of the advantages of the adaptive large neighborhood search algorithm,indicating the effectiveness of the proposed insertion operator and adaptive mechanism.Finally,the distribution mode proposed in this thesis is simulated with the actual distribution data of S company.The simulation results show that the distribution distance of the distribution mode in this paper is much shorter than the actual distribution distance of S company.To sum up,this thesis proposes an open multi-trip two-echelon vehicle routing problem in which LV can carry SUAV,and designs two algorithms to solve the problems with and without time window constraints.These provide theoretical support and decision support for building modern logistics systems and hence,it has important theoretical and practical significance...