Research On Vehicle Travel Time Prediction Based On Bus Data

With the expansion of city,the public demand for travel is increasing.It becomes a hotpoint recently to estimate the travel time using the mass data which generated by the intelligent transportation system.In this paper,bus data are analyzed and calculated by Multifractal,Detrended Cross-Correlation Analysis,Radial Basis Function(RBF)Neu-ral Network and Vehicle Travel Time Prediction methods to research vehicle travel time prediction.Which has a reference value to improve the quality of urban transport servi-ces.In this paper,we deeply analyze the domestic and foreign research present situ-ation,using Multifractal Detrended Cross-Correlation Analysis method,analyze vehicle data,and the results prove that in given time range,within the same multifractal range,speed of buses and vehicles to be predicted showe a positive corr-elation.Base on the correlation research,we use the improved RBF neural network to build a TSMSA-RBF Vehicle Speed prediction model.This model build RBF-NN off-line firstly,and train it online by the bus data,and Scence Selector is used to change network configuration to self-adapt different scences.Through the experiments and verification in a large number of actual data,it is tested that the model works well in speed prediction.Based on this model,using the online map API and Vehicle Travel Time Prediction method,Vehicle Travel Time Prediction simulation is implemented in the paper.It could simulate the real vehicle driving,and show the on-time information of driving,then calculate the total travel time on the route.Vehicle Travel Time Prediction program is implemented based on the research on theory,which achieve the goal of this paper.And it is a useful exploration for further re-search on the comprehensive intelligent transportation,and provide a reference for en-hance the quality of transportation services and public travel convenience.