Research On Recognition And Forecast Of Traffic Congestion Based On Online Car-hailing GPS Data

With the rapid development of the Internet,online car-hailing service has gradually entered people’s lives and brought great convenience.However,traffic congestion has always been a major and difficult problem in urban traffic.If you can identify the traffic status of vehicles on the urban road network and make short-term traffic predictions,it will greatly improve the convenience and comfort of travel.This article takes the trajectory data of more than 50,000 vehicles in the second ring road of Chengdu in November 2016 as the basis for research.First,perform preprocessing operations on the trajectory data,including data cleaning,data coordinate conversion,map matching,and trajectory data segmentation.Second,feature analysis of trajectory data,including classification of traffic status,identification of traffic parameters,analysis of spatiotemporal characteristics of trajectory data,screening of trajectory data,calculation of indicators and analysis of travel characteristics,etc.Among the travel characteristics of online car-hailing,there are three peaks for travel at the pick-up points,the morning peak(8:00-10:00),the afternoon peak(12:00-14:00)and the evening peak(17:00-19:00),the travel time is mostly concentrated in 5-25 minutes;in addition,in terms of spatial distribution,it is different from cruise taxis that are distributed around the main road or near residential areaFinally,combined with the actual case of Chengdu,the traffic congestion prediction research is conducted in different time periods and different sections.The one-month data is divided into two parts: a training set and a test set.Combined with POI data and extracted parameters,a set of road traffic state influencing factors is constructed.As a model training,the Hidden Markov prediction model is used to conduct short-term road traffic congestion.Forecast,and finally use the test set to verify.The results of the study show that the accuracy of the model has a close relationship with the study area and evaluation index parameters.Taking the goodness-of-fit evaluation index as an example,in terms of the accuracy of the prediction results,the prediction effect in the main urban area is better than in other research areas.