Demand Forecasting And Location And Routing Planning Of Front Warehouse For Company D

With the development of network technology and the increase of consumer demand,the transaction scale of fresh e-commerce platform in China has increased from less than 50 billion yuan to nearly 450 billion yuan,with the transaction scale increasing to 27.92% year-on-year and the market penetration rate reaching 7.91%.As a popular business model in recent years,the front warehouse mode has been popular with many fresh e-commerce platforms,which have laid out thousands of stores in large cities.However,in the current warehousing mode,when the fresh e-commerce enterprises are expanding in scale and seizing the market in full swing,due to the incorrect estimation of the target market demand and unreasonable layout of the front warehouse,most platforms are not profitable,even with serious losses,and eventually withdraw from the market,resulting in a large waste of resources.Based on the above background,this paper studies the operation mode of D Fresh E-commerce Company,summarizes the cost problems faced by its front warehouse mode,and through predicting the demand of users in the target market,further studies the location and path planning of the front warehouse.The work of this paper is mainly composed of five aspects:(1)Problem Analysis of D Company’s Front Warehouse Mode.This paper first gives a detailed introduction to the operation mode of Company D,and at the same time,makes a cost analysis of its front warehouse mode.Finally,the theoretical minimum order density under the front warehouse mode of Company D in Chengdu is obtained on the basis of maintaining the profit-loss balance.It lays the data foundation for the subsequent layout of the front warehouse.(2)D Company’s User Demand Forecast.Since D Company expands its market to Chengdu without historical data of order volume,this paper solves the problem by using BP network model with strong non-linear mapping ability and flexible network structure and GM(1,1)gray prediction model with high accuracy for short-term data prediction.First,the impact indicators are filtered and trained on the existing order data in Shanghai,and a BP network prediction model is established.At the same time,the GM(1,1)gray prediction model is used to make short-term prediction of the relevant predictive influencing factors in Chengdu,and the relevant indicators from2022 to 2024 are obtained.Finally,the numerical value is substituted into the BP network prediction model that has been trained to predict the order volume in Chengdu from 2022 to 2024,and the prediction results are obtained.(3)D Company’s front warehouse location.This paper considers the layout planning of city-wide front warehouse,so no alternative points are set.This problem has a large range and a large number of features,so a more suitable clustering theory is used to solve the problem.This paper is divided into three steps: the first step,to initially determine the demand points.Based on population density,ARC GIS software was used to establish a 1 km*1 km raster within the city as user demand point,and then preliminary screening was carried out according to population density.The second step is to cluster for the first time.Based on the calculation results of the minimum order density of the front warehouse mode determined in Chapter 2,an improved DBSCAN algorithm using the sum of orders in range as the density threshold is used to exclude the customer points that do not meet the density requirements and obtain the first clustering result.The third step,the second clustering.Based on the first clustering result,a quadratic clustering is performed on the clustered areas that exceed the front warehouse capacity and extent,and an improved K-Means algorithm with distance/demand as the compactness of the clustering is used.Finally,the coordinate location and service range of each front warehouse are obtained.(4)D Company’s distribution route planning.This paper establishes a multi-objective vehicle routing problem with time windows for the center warehouse-to-front warehouse distribution route planning problem of fresh e-commerce company D.The objective function is to minimize the total cost and maximize the freshness of fresh products,while increasing the penalty cost for the time window that precedes or exceeds the replenishment time of the front warehouse.NSGA-II algorithm is used to solve the problem and the Pareto optimal solution set is obtained.This paper has completed the analysis of the front warehouse model of Company D,the forecast of the target market demand,the location of the front warehouse and the distribution path planning.On the one hand,it provides a reference for the logistics cost expected to develop a new market and carry out the layout of the front warehouse;On the other hand,according to the demand forecast results,a scientific,reasonable and practical decision-making scheme for the layout of the front warehouse and its distribution path is proposed.When the company will further expand the market in the future,it can ensure the cost and efficiency of the logistics system of the front warehouse and improve its market competitiveness.