Research On The Vehicle Routing Problem Of The Mixed Fleet Of Autonomous Vehicles And Conventional Vehicles Considering Distribution Mode

At present,energy shortage,serious environmental pollution,traffic congestion,frequent traffic accidents and other global problems have spawned the birth and development of autonomous vehicles(AVs).Compared with the high-speed and manned vehicles in transportation,low-speed and unmanned truck is the best application area for the existing AVs,that is,unmanned distribution.From the birth of the concept of unmanned distribution to the large-scale application,it takes a long transitional phase.In this phase,due to the immature technology and high cost of the autonomous vehicles,it is the feasible and inevitable way for companys to partially replace conventional vehicles(CVs).Therefore,it is theoretically and practically significant to research on the vehicle routing problem of the mixed fleet of AVs and CVs based on distribution mode.Firstly,the study of the unmanned distribution is carried out.From the aspects of restricted general policy,strict delivery time,distribution vehicles,operation costs and scheduling mechanism,the differences between the unmanned and the manned distribution are clarified.Meanwhile,corresponding to the development stage of AVs,three development stages of the unmanned distribution are put forward: test,transition and large-scale application phase.For the transition phase,three distribution modes of the mixed fleet of AVs and CVs are proposed: zoned distribution,mixed distribution and classified distribution.Secondly,the research of the autonomous vehicle routing problem is carried out.Considering the path interruption,restrictions of load capacity and delivery time,the AVRPTW model is established.And an improved PSO algorithm is put forward to solve the AVRPTW model.By numerical study,AVRPTW model and the improved algorithm are proved to be scientific.Thirdly,based on the mixed distribution mode,the vehicle routing problem of the mixed fleet of AVs and CVs is studied.Considering the fact that AVs and CVs operate in the same region,but AVs' operation are restricted by path interruption and delivery time,the MTVRPTW model is established.And a 2-opt PSO algorithm is proposed to slove the problem.By numerical study,the 2-opt PSO algorithm is proved to be efficient to solve the MT-VRPTW model.Besides,simulation results show that increasing load capacity and removing the time windows of AVs can effectively promote the application of AVs and reduce costs.Lastly,research on the vehicle routing problem of the mixed fleet of AVs and CVs based on the classified mode is carried out.Considering a secondary logistics network system,AVs and CVs are classified to perform the delivery tasks.A 2E-VRPWT model and a corresponding two-phase solution are proposed.The customer is grouped by the principle of minimum distance distribution,and the 2E-VRPTW model is transformed into the lower CAVRPTW and the upper CDVRP model.According to the characteristics of the model,genetic hybrid simulated annealing algorithm is designed to solve the models.Based on the true network data from ‘Cainiao Network Technology' company,a case study is carried out.The solution idea and algorithm are proved to be efficient to solve the proposed 2E-VRPTW model.