Research On Volume Prediction And Route Optimization Of H Company's Container Dumping Transportation

With the rapid development of economy,people's demand for logistics is growing.As a result,the distribution of the receivers is more dispersed,and the volume of most receivers is less,so the volume of a single receiver is difficult to reach the volume required by vehicle transportation.How to reduce cost and improve efficiency,and provide better service and user experience for customers through container dumping transportation and scientific path optimization is one of the problems that logistics enterprises need to solve.As a logistics enterprise,H company's container dumping transportation business is one of its main businesses.Due to the time lag of container dumping volume statistics,in order to optimize the container dumping transportation path between H company's internal port station and distribution center.In this paper,we first find out the appropriate forecasting method to forecast the cargo volume of each station of H company.Then,taking the predicted value of each station as the initial value of the path optimization model,the transportation path planning is carried out,and the initial transportation scheme is obtained.Finally,according to the actual cargo volume of each station,the initial transportation scheme is dynamically adjusted to achieve the goal of the shortest overall transportation path of H company.Since the statistical unit of the volume of dumped containers is containers,in order to minimize the cumulative error caused by the upward rounding sum,this paper proposes a two-stage prediction model of the volume of each station.The first stage:total cargo forecasting model.In order to find a better fitting way to forecast the total volume of goods.In this paper,two stations are randomly selected from 21 stations as samples,and the firstorder linear fitting,second-order linear fitting and BP neural network rolling prediction are used for the prediction of the two samples respectively.By comparing the goodness of fit of the three methods,it is concluded that BP neural network rolling prediction is a relatively better prediction method.Therefore,this method is used to predict the total volume of 21 stations.In the second stage,based on the 0-1 knapsack of each station volume prediction model.Each site is regarded as an article,the proportion coefficient of 11 day volume of each site in the total 11 day volume of all sites is regarded as the value coefficient of the article,and the proportion coefficient of single day maximum volume of each site in the total maximum value of all sites is regarded as the volume coefficient of the article.The 0-1 knapsack method is used to allocate the total volume forecast value to each station,so as to obtain the volume forecast value of each station.In this paper,the transportation route optimization problem of H company is transformed into CVRP problem,and an improved SA algorithm is proposed to solve this problem.The method of generating new solution is set to select by roulette rules,and one of the three methods of interchange,reverse order,and random insertion is used to generate new solution.The transportation path of H company is optimized.Finally,this paper gives some suggestions on the optimization of H company's container dumping transportation.