Research on a Highway Passenger Volume Prediction Model Based on a Multilayer Perceptron Neural Network

Abstract

The accurate prediction of highway passenger volume is very important for China’s transportation planning and economic development. Based on a neural network, this paper establishes a prediction model by using historical road passenger traffic and related influencing factor data, aiming to provide an accurate road passenger traffic prediction. Firstly, the historical highway passenger volume data and the factor data affecting passenger volume are collected. Then, a multilayer perceptron neural network is established by using SPSS software (PASW Statistics 18) to analyze the significant relationship between highway passenger volume and influencing factors. Then, through the training and verification of the model by MATLAB software (R2021a), the reliability of the prediction model is proved. Finally, the model is used to predict the passenger traffic volume in 2020–2022, and the actual passenger traffic volume is compared and analyzed. It is concluded that the highway passenger traffic volume decreased significantly in 2020–2022 due to various factors such as the epidemic situation and policies, which have had an impact on China’s economic development.