Prediction Of Highway Freight Volume In Sichuan Province Based On Wavelet Packet Decomposition And Neural Network

With the acceleration of the construction of highway transportation hubs and logistics and freight facilities,Sichuan’s logistics industry is developing rapidly in all directions,and the level of highway freight is constantly improving.The volume of road freight in Sichuan Province is the main indicator to measure the development of the transportation industry in Sichuan Province,an important basis for reflecting the economic development of Sichuan Province,and an important reference for the government to formulate road transportation infrastructure plans.In order to rationally plan the construction of road transportation infrastructure in Sichuan Province and balance the supply and demand of road cargo transportation,it is necessary to accurately predict the amount of road transportation in Sichuan Province.At present,there are many forecasting methods for road freight volume,but they all have various problems.First,the original road freight volume data contains random fluctuation data,which affects the accuracy of freight volume forecasting.Secondly,there are many factors affecting road freight volume forecasting,and it is difficult for general forecasting methods to deal with such non-linear and multi-data freight volume forecasting problems.In this context,how to improve the prediction accuracy of highway freight volume is of great significance to the planning of highway transportation infrastructure and the development strategy of the logistics industry in Sichuan Province.In view of this,this thesis conducts research on the forecast of highway freight volume in Sichuan Province.First,it sorts out the research results of domestic and foreign road freight volume forecasting,and summarizes the current research status.Then introduce the theoretical models related to the highway freight volume used in this thesis,mainly introduce the gray prediction model,the BP neural network model in the neural network model,the long short-term memory(Long Short-Term Memory,LSTM)network model,and the wavelet packet decomposition theory(Wavelet Packet Decomposition,WPD).Then,analyze the influencing factors of the highway freight volume in Sichuan Province,based on Pearson’s correlation analysis,according to the actual traffic and transportation conditions in Sichuan Province,follow the factor selection principle,and get the main influencing factors of the highway freight volume in Sichuan Province.Next,the construction of a road freight volume forecasting model in Sichuan Province,following the principles of forecasting models,is the first to propose a road freight volume model based on the WPD-based LSTM network(WPD-LSTM model),formulating forecasting steps,and designing evaluation indicators for forecast results.Finally,collect and sort out the historical data of Sichuan Province’s highway freight volume and related influencing factors,and conduct an empirical analysis of Sichuan Province’s highway freight volume forecast.This paper uses WPD to process road freight volume data,trains and constructs WPD-BP neural network model and WPD-LSTM network model,and predicts highway freight volume in Sichuan Province.Then,the gray model,BP neural network model and LSTM network model are used to conduct comparative experiments with the prediction model proposed in this paper,and the prediction results of the five models are compared and analyzed with the real results.The experimental results show that WPD can effectively improve the prediction accuracy of highway freight volume.The WPD-LSTM neural network model performs well and can predict the highway freight volume of Sichuan Province well.The method proposed in the thesis is feasible and accurate.