Research On The Prediction Method Of Bus Travel Time In The Network-connected Traffic Environment

With the increasing ownership of vehicles,traffic congestion or become a major problem of urban sustainable development,public transport priority development is one of the important means to alleviate urban traffic congestion problem.Road bus vehicle by society and the diversity of intersection signal control,travel,make the bus driving condition is uneven,lead to unstable bus travel time,affect the passenger bus travel experience.Study of bus travel time prediction method,to improve the service level,improve public transit bus attraction is of great significance.Papers on domestic and foreign scholars in recent years based on the analysis of the research results,on the bus lanes scenarios,bus station time prediction error,problems such as instability,passengers waiting time for bus travel time prediction method made the traffic environment study,mainly includes the following several aspects:(1)For complex drive mode existing bus data usually lack of vehicle in and out of the station process information,and online maps using only a single point is the location of the station,the first bus station time influence factor analysis,based on DNN bus station time prediction model is constructed;And prediction error set on the child section division,for each section USES the nonparametric kernel density estimation algorithm,build up the probability interval prediction model,for a certain confidence level under the bus station to predict the probability interval time.Through the analysis of the forecast results,this paper carry out of the station based on DNN-NKDE time probability interval prediction method in combination algorithm of correlation between the predicted values and the true value than forecast model is based on the DNN bus station for the 12%.(2)For bus data cannot accurately reflect the dynamic traffic condition and lead to the low prediction precision,using simulation software to build road network access snatched SUMO vehicle track data,and the track of time and space in structure analysis,based on course length constraint and uniformity of trajectory segmentation method,the trajectory,multiple sub divided into based on considering the headway and travel time trajectory matching hidden markov model,on the same road on the same time the highest motion similarity in line trajectory matching implementation approaches to hand in the station between travel time prediction.Finally,integrating the bus station time prediction of travel time between prediction and stand,get the bus in a path interval travel time prediction,contrast analysis shows that the proposed snatched the traffic environment of bus travel time prediction method is effective,and has better predict reliability.