Research On Urban Road Short-Term Travel Time Prediction Method

With the growing economy and the rapid development of urban vehicles,road network encountered many problems,such as traffic congestion,air pollution and so on.Traffic has become an urgent problem in major cities.Intelligent Transportation Systems(Intelligent Transportation System,ITS)is an effective way to solve the traffic problems.Traffic guidance system,as an important part of the intelligent transportation system,plays a big role in improving road congestion,reducing accidents and air pollution.Vehicle travel time is an important parameter in the traffic guidance,which reflects the urban road traffic conditions.Therefore,it is nessasary to study the the travel time prediction of urban road.This paper discusses the urban road network in the short-term travel time prediction.The main work includes:According to the characteristics of urban road network,the significance of travel time prediction technique is described.Through traffic characteristics,the principle and characteristics of the existing vehicle travel time prediction technique is discussed.Short-term travel time prediction algorithm Based on Extreme Learning Machine(ELM)has been proposed.Based on the characteristics of nonlinear and time-varying traffic parameters and others characters,Extreme Learning Machine has been chosen as the travel time prediction model for its self-learning machine learning,fast good characteristics.Simulation experiments is predicted on the REGIOLAB-DELFT platform by using real traffic data.Results show that in the sample and sample fluctuation ELM adequate travel time prediction model under two scenarios show a higher accuracy and applicability by comparing the ELM and support vector regression(SVR)and BP neural network(BPNN)comparing experimental.On the basis of integrated Bagging and ELM,BG-ELM algorithm has been proposed to prediction short-term travel time.Through the analysis of experimental results,it is found that in peak periods with higher volatility data,ELM errors are bigger.Therefore,in order to improve the ELM generalization ability,bagging algorithm is introduced to train multiple ELM,and the obtained multiple results are synthesized to improve ELM generalization ability.Experiment results show that the BG-ELM shows higher stability and accuracy than ELM in travel time prediction.