Sales Forecast Of New Energy Vehicles Based On Improved LSTM And Online Review Mining

Cars are now an indispensable means of transportation in life.It can be said that cars have penetrated into every corner of life and promoted the development of social economy.However,due to the increasing loss of energy and the increasing pollution of nature,the challenges facing the earth are increasing day by day.As of June 2020,the number of cars in my country has reached 270 million,which has brought enormous pressure to the ecological environment.In recent years,low-carbon and environmentally friendly new energy vehicles are slowly changing this era.The advocacy and support of the state,as well as the energy conservation and emission reduction requirements of large and small enterprises,are promoting the rapid development of new energy vehicles.Under normal circumstances,the sales volume of the automobile market can be estimated through the traditional vehicle sales index estimation method.However,the current new energy vehicle series is rapidly innovating,there is no complete market sales management system,and the market sales data that can be analyzed are also relatively low.It is more difficult to estimate its sales volume.At present,the key research object of automobile sales forecast is still traditional gasoline models.As a vehicle category supported by the state and society in recent years,new energy vehicles not only play a great role in environmental protection and air quality improvement,but also because of industrial characteristics,cash flow Due to reasons such as small size and small size,it is more likely to encounter the situation that production and sales are out of sync,resulting in huge loss of profits,and even affecting the development of new energy vehicles and research on new technologies.The analysis and forecast of the market sales of the new energy vehicle industry can help clean energy vehicle manufacturing companies gain a favorable position in the domestic and even global market competition,effectively predict the sales of new energy vehicles,and ensure that enterprises can produce in a short time.The reliability and robustness of the plan,and through the rational allocation of various resources and other means,can effectively further optimize the company’s inventory.Through data verification and analysis,the following conclusions are drawn:(1)The new energy vehicle sales data studied in this thesis are all short-time series data.After data preprocessing and feature analysis,LSTM neural network is used to analyze and predict the short-time series.And through the Maxout activation function method to optimize and improve the activation function structure of the LSTM neural network,the experimental results show that both the original LSTM neural network and the improved ones are better than the SVR and gray prediction models,and the Maxout method improved The prediction accuracy of LSTM is the best.(2)The model combines monthly car sales data with market evaluation data,and designs and completes a sales forecast model based on review feature expansion based on the improved LSTM sales forecast model.Experiments show that this model has a relatively high accuracy,and can lay a foundation for future research.