The Study On Real Time Prediction Of Bus Arrival Time In Cities

In recent years,Public transit-oriented urban traffic has been developing quickly,and the Advanced public transport system(ASPT)attracts more and more attention of city dwellers.At the same time,the real-time prediction of bus arrival time has also become a research hotspot.With the real-time and accurate prediction technology of bus arrival time,it helps promote the development of APTS intelligent information,and improve the management level and service ability of mega cities,and improve the operation efficiency of public transport system at the same time.Therefore,this paper selects the bus routes in Guangzhou as the research object,and carries out research on the real-time prediction of bus arrival time of large city.We analyzed the history time data of bus running,and then constructed the running time curve.We stated from the statistical law,and put forward the prediction model of curve matching based on clustering analysis.The main research content is divided into the following three aspects:First,the research based on the data of Guangzhou bus route in recent 110 days in 2017,then we make the raw data get preprocessed and standardized analysis.The historical data were transformed into visualized running time curves.The curves are processed by median filtering in order to study the difference of the running time curve.We put forward a research idea based on curve matching aim to curve characteristic parameters,such as slope,ordinate value,curvature and so on.Then,in order to study the efficiency and accuracy of the bus arrival prediction based on the curve matching.We start from the matching of the characteristic parameters of the bus arrival time curve,and the K-means clustering analysis is selected for the curve of the last hundred days,according to the clustering speed and the characteristics of the variable type.A number of runtime curves were sorted to the best clustering method,so that the practicability and accuracy of the deep level test curve matching were achieved.Finally,according to the real-time prediction model of the bus arrival based on the curve matching of clustering analysis,the historical data of a certain line in Guangzhou is tested.First,the original data are perfected and standardized,and then we construct the real-time prediction model of bus arrival.By comparing the accuracy and efficiency of the three prediction results,such as the prediction model based on the date feature,the curve matching prediction model based on the cluster analysis and the curve matching prediction model,we find the The curve matching prediction model is effective in two aspects of predi ction accuracy and computational complexity.In summary,in view of the low efficiency and low accuracy of the existing bus arrival real-time prediction model,this paper presents a curve matching prediction model based on cluster analysis,and more consid erations on the factors such as the early peak,date characteristics and traffic volume of the megacities,which provide the data for the construction and management of the advanced public transportation system.And theoretical support..