Design Of Sales System Based On Sales Forecast Model

In 2020,due to the impact of the epidemic,people avoid face-to-face shopping,which makes automatic unmanned vending machines such as automatic vegetable machines and automatic beverage machines more popular in the market.The vending machine market has also ushered in another peak of development.Traditional automatic vending systems have fallen behind in terms of payment methods,state monitoring methods of vending machines,and so on.The problem of vending machine replenishment and the ratio of regional replenishment personnel has become a difficult point in the industry.This problem directly affects the profitability of merchants.To solve this problem,it is necessary to establish a sales volume prediction model and design in advance based on the prediction results of the model.Replenishment frequency and replenishment personnel ratio plan.This project relies on the existing cloud platform of the Internet of Things,combined with the MQTT protocol and 4G module to design a sales shopping system,and then save the order data in the cloud platform through the My SQL database.The historical sales data in the database is exported,four sales prediction models were used to predict the sales volume of vending machines.It can be seen from the experimental results that the prediction error of the combined model based on the reciprocal error method is significantly smaller than that of the other three models.The model provides decision basis for replenishment plan,thereby saving labor costs and increasing sales,and finally maximizing the revenue of the operator.The specific work of this paper is as follows:(1)Introduced the development status and research hotspots of vending machines,and determined the research objectives: First,the data transmission between the vending machine terminal and the cloud platform meets the characteristics of high reliability and low traffic.Second,establish a sales forecast model to obtain an optimal replenishment plan and improve economic efficiency.(2)Starting from the research requirements of the vending system,Based on the design architecture of the sales system,the technical architecture analysis is carried out,Then the hardware design of the vending device and the corresponding driver configuration process are illustrated,and the software realization process of the device is explained in detail(3)Analyzed the influencing factors of vending machine sales,expounded the construction of sales order database and data sources.Finally,a sales sample data set was established.Theoretically analyzed the feasibility of ARIMA model and support vector regression model(SVR)in sales forecasting,expounded the modeling process of the two methods,and used these two methods to predict sales.The results show that the ARIMA model has the problem of ignoring nonlinear influencing factors,but the weight of nonlinear influencing factors is greater than the historical time factor,which leads to the larger prediction error of the ARIMA model in this subject.Based on the previous experimental results,taking into account the data mining capabilities of the deep neural network(DNN)under non-linear factors,and according to the modeling process of the deep neural network,DNN is used to predict the sales volume.Based on the good forecasting effects of SVR and DNN,a linear weighted combined forecasting model(SVR-DNN)is used to predict the sales volume,in which the weight coefficient is determined by the variable weight error reciprocal method.The results show that the comprehensive forecasting performance of the combined forecasting model is significantly better than that of other single models.(4)Test the stability of the vending machine device,test the accuracy of the data communication between the vending machine terminal and the cloud platform,and then analyze and compare the four models from various dimensions through experiments,verifying the automatic design of the SVR-DNN prediction model in this subject The superiority of vending system application.Finally,the reliability of the sales system designed by this subject is proved.