Research On New Energy Vehicle Sales Prediction Based On Neural Network

In recent years,China new energy vehicle market has developed rapidly,and the number of consumers buying new energy vehicles has increased year by year.Since 2010,China has gradually implemented the new energy vehicle industry,and at the same time has issued a series of relevant policies to support the development of it..With the gradual expansion and development of the market,there are some star products that can achieve double harvests in sales and profits,and there are also some products that have been on the market for a long time but sold not well..Based on this,this paper takes BYD’s new energy vehicles as an example,and hopes to build a model based on the domestic new energy vehicle market in order to achieve the forecast of future sales of different new energy vehicles in the domestic market environment,by studying the sales of its new energy vehicles in recent years and deeply analyzing the factors affecting the sales of its different vehicles.In the previous studies on the domestic new energy vehicle market by other scholars,most of them discussed the prediction model of the overall sales of new energy vehicles from the perspective of macroeconomics rather than several specific companies and their vehicles.The focus of this study is to determine which vehicle parameters,national policies and vehicle environment factors consumers pay more attention to when purchasing new energy vehicles,through online information researching,and select the more critical elements of the above factors as independent variables.Take the cumulative sales volume of the vehicle of the third month in the future is used as a dependent variable to discuss how to build a sales forecast model based on a multi-layer sensor neural network.This paper will determine BYD as the research object through data research,and download some of BYD’s new energy vehicles’ data from Internet searches as specific research samples,to demonstrate the feasibility of predicting future sales of models through data such as pricing,vehicle parameters,national policies,and vehicle using environment.In general,this paper has carried out several aspects of work as follows:(1)Selected 20 BYD new energy vehicles,and carried out descriptive statistical analysis and cluster analysis on their parameters and sales,which proved to a certain extent that the parameters of BYD new energy vehicles can affect their sales.(2)The sales forecast model is established by using multilayer perceptron neural network and multiple regression mode.Among them,the prediction model established by the multi-layer perceptron neural network has higher accuracy but greater volatility.The prediction model established by the multiple regression model has relatively small volatility but poor accuracy.(3)Using the support vector regression algorithm,on the basis of the multilayer perceptron neural network,combined with the multiple regression model,a combined sales forecast model was established,which realized the optimization of both accuracy and volatility.The established combined forecasting model can predict the sales of multiple new energy vehicle under the same brand for a period of time in the future,thereby helping enterprises determine research and development priorities and arrange production plans.At the same time,this paper proposes ideas for optimization and improvement of the final model,which will serve as a reference for future research in related fields.