Unmanned Delivery Vehicles Assisted Truck Distribution Route Optimization Under Time-Dependent Road Networks

With the rapid development of the logistics industry and the advent of the age of aging population,the labor cost of the logistics industry has gradually risen,and the concept of unmanned distribution has become popular.In recent years,unmanned delivery vehicles have gradually came into public view,and organized small-scale distribution has been carried out in campuses,factories and residential areas.However,unmanned delivery vehicles cannot achieve all-area autonomous distribution for the time being,and the trucks cannot effectively solve the "last kilometer" due to the distribution area restrictions.Therefore,the unmanned delivery vehicle assisted truck distribution mode emerges.In this mode,the unmanned delivery vehicle follows the truck in the way of pick-up.When it needs to deliver customers,the unmanned delivery vehicles and trucks are separated and distributed respectively.After completing the service,the unmanned delivery vehicles arrive at the confluence point and join the trucks.This distribution mode effectively improves the distribution efficiency and shortens the distribution time.At present,the traffic environment of the research literature on the distribution of unmanned delivery vehicles and trucks is mostly static road network,but in reality,the vehicle speed is dynamically changing.The research results can be closer to the reality by considering the time-dependent road network.This thesis studies unmanned delivery vehicles assisted truck distribution route optimization under time-dependent road networks.The main research contents are as follows:Firstly,this thesis chooses the appropriate time-dependent description method and establishes the time-dependent function model.By analyzing the factors such as vehicle load,customer demand and regional restriction,a mixed integer programming model with the goal of minimizing distribution time was constructed.Secondly,the solution process is divided into two parts.The first part is the generation of initial solutions.In order to improve the algorithm performance,K-means clustering algorithm is firstly used to cluster customers,and then the nearest neighbor strategy of greedy algorithm is used to generate initial solutions.In the second part,five neighborhood operations,such as cross-insert operation,global exchange operation and greedy insert operation,are designed.The scores,weights and probabilities of the neighborhood operations used are updated by the adaptive adjustment mechanism,and the solution acceptance strategy is formulated by combining the Metropolis criterion in the simulated annealing algorithm,to obtain the adaptive large neighborhood search algorithm designed in this thesis.Finally,the adaptive large neighborhood search algorithm was used to solve the Solomon data set of different scale examples,and three sets of experiments were designed to verify that the algorithm is effective and feasible for solving the distribution route optimization problem of unmanned delivery vehicles assisted trucks.In the first experiment,it is concluded that the higher the precision of time period division,the better the experimental effect,by changing the time interval division accuracy.The second experiment is to compare the distribution strategies under the static network and the time-dependent road network,and draw the conclusion that it is necessary to consider the time-dependent road network in the unmanned delivery vehicles assisted truck distribution route optimization under time-dependent road networks.The third experiment is to change the departure time,and draw the conclusion that logistics enterprises should flexibly arrange the departure time within the distribution working time to reduce the distribution time.