Study On Prediction Method Of Bus Arrival Time Of Urban Public Transport

With the continuous development of the national economy,urban population continues to increase,and the per capita car ownership continues to increase,resulting in a continuous increase in the number of urban motor vehicles.However,the construction speed and scale of the existing transportation infrastructure can not meet the increasing demands of urban transportation,and the contradiction between supply and demand is prominent,making the urban traffic increasingly congested.To solve urban transport problems,urban public transport needs to be developed.At this stage,in many small and medium-sized cities,conventional buses are the main component of public transportation.Conventional buses are the best way to solve the problem of urban residents traveling with the widest coverage,the most running routes,and the lowest bus fares.However,at this stage in many small and medium-sized cities,the proportion of conventional public transportation is not high,and public transportation is less attractive to residents.The main reason is mainly due to the uncertainty of the regular bus arrival time,poor punctuality,passengers need to wait for unknown time,need to pay attention to the arrival of vehicle information,public transport information release is relatively backward,the emergence of 'neck' and other public transport situation.Passengers are prone to anxious waiting,or directly change to other modes of transport.Therefore,accurate real-time bus arrival time prediction can improve the proportion of small and medium-sized urban residents bus travel,improve passenger satisfaction,improve the service level of urban public transport,is of great significance to solve the traffic problem.This paper first analyzes the operating characteristics and influencing factors of bus arrival time,and divides the city bus arrival time into three parts,namely,the road travel time,the stop time and the intersection delay time.Based on the operation characteristics and influencing factors of these three parts,the factors of bus-to-station time prediction are selected.For the GPS data produced by bus operation,the algorithm of data processing and the algorithm of linear matching of bus are designed to obtain the bus operation data.Through the analysis and study of several bus-to-station time prediction theories commonly used at home and abroad,the model established in this paper has laid a theoretical foundation.The model of this paper is proposed,first of all,on the basis of obtaining a certain amount of bus arrival time analysis and data,use bus vehicle GPS historical information and real-time information,to process it,and then obtain the historical arrival time of bus operation and real-time running speed and vehicle location and other information.The third step,using the advantages of high precision in short-term prediction,uses it for the prediction of bus's next stop to station time,establishes a real-time prediction model based on the bus GPS information arrival time based on the previous work,and finally uses the real-time time time and prediction time of the bus to the last station as the input of Kalman filter,and uses Kalman filter ingester equation to predict the bus to the station time,so as to establish the real-time bus to the time forecast based on the dynamic bus to the station based on the calm filter.In order to verify the proposed model,this paper selects three representative bus lines in Almaty city to carry out the model of the example of the summary certificate.By collecting GPS data generated by the 11-day operation of buses on Almaty 92,60 and 44,and by the actual operation of buses on the three lines,the stations were actually running to the time to collect data from the bus.The obtained GPS data is processed according to the method introduced in this article,and the time information,speed information,location information and so on of the bus operation are obtained.The acquisition example verifies the bus-to-station time prediction model built in this paper,and uses the average absolute error and the average absolute error percentage as the index of error analysis.By comparing the model prediction results and the actual values,the results are evaluated by using the error evaluation index,and it is concluded that on these three different characteristic strains,the bus-to-station time prediction,whether peak or non-peak,has a more accurate prediction effect,the accuracy of the prediction is more than 90%.