Research On Location-routing Problem Of Fresh Agricultural Products Origin Distribution Center

With the continuous improvement of agricultural production technology and economic development level,China’s agricultural production has maintained a steady growth,China has also become a major agricultural producer in the world.However,due to various reasons,the loss of agricultural products in the process of transportation is huge,and the annual rotten fresh agricultural products reach hundreds of millions of tons.Among them,the transportation process of goods collection at the production end has a direct impact on the subsequent trunk transportation and terminal distribution process,But at present,fresh agricultural products directly enter the circulation market without pre cooling after picking in the producing area,and the unreasonable allocation of the distribution center leads to the increase of transportation cost.In order to improve the node layout of agricultural products origin,optimize the route network from the origin to the distribution center,and reduce the cargo loss in the process,This paper studies the location routing problem of agricultural products origin distribution center.At present,China’s agricultural production is still dominated by decentralized small-scale production,taking into account the large output of general producing areas This paper establishes a location path model considering the split demand,and takes the minimum total cost of transportation cost,cargo loss cost and refrigeration cost as the optimization objective.In the aspect of model solving,this paper divides the solving process into two stages,In the first stage,the heuristic algorithm is used to determine the selected facilities and divide the customer demand points,that is to solve the location allocation problem.In the second stage,deep reinforcement learning is used to plan the vehicle routing for the demand points served by each facility.In this paper,deep reinforcement learning algorithm is applied to solve the vehicle routing problem to verify its superiority in solving large-scale problems.The main work of this paper is as follows: firstly,the paper makes a basic study on location path problem,vehicle path problem with split demand and deep reinforcement learning.Then,the focus and research direction of current research are visually analyzed by using the literature measuring tool Vos viewer.Secondly,based on the analysis of various costs,the location path model of fresh agricultural products distribution center was established with the objective of minimizing the total cost.Considering whether the demand can be split into two modes,two models are established and solved by two-stage algorithm.Finally,combined with the example of S company,the problem is solved and analyzed.Through the comparison of two different modes and algorithms,the following conclusions are drawn:(1)Considering the demand split mode can reduce the number of vehicles used,and can greatly shorten the driving distance of vehicles,reduce the cost of goods collection process,improve the efficiency of goods collection.(2)For large-scale vehicle routing problem with more nodes,deep reinforcement learning can achieve better results than general heuristic algorithm in solving effect and running speed.(3)Through the analysis of the parameters,the importance of precooling in producing area is proved to a certain extent,which can greatly reduce the product loss in the transportation process and reduce the loss.