Research And Implementation Of Multi-factor Hybrid Vehicle Flow Prediction Model

With the continuous improvement of the economic level,residents'requirements for the quality of travel have gradually increased.However,traffic jam,traffic safety and other issues are still the main factors restricting urban traffic development.If we can make an accurate forecast of short-term volume of vehicle,especially during holidays,bad weather and accidents,we will provide the basis for the urban traffic department and it can take traffic dredge and congestion response measures earlier.It is helpful to improve the level of urban traffic information management.This project is based on data from the traffic department,researching the characteristics of urban volume of vehicle changing with various factors.The method of combining the autoregressive ARIMA model in mathematical statistics with the CART model tree in machine learning is proposed and used for the prediction of short-term traffic flow.It has been verified that this method can have better predictive performance when an emergency occurs.After completing the experiment and verification of the combined model for volume prediction of vehicle,this project also designed and implemented a mobile terminal prototype system for traffic prediction and warning.The system uses a popular hybrid development model,basing on the Cordova framework and API of AMAP,and the back-end uses the python language for data calculation and storage.Through this model and system,it can provide timely forecasting and warning information for the traffic department and improve the emergency response capacity of urban traffic in the face of sudden traffic congestion.Finally it can improve the travel quality of citizens.