Research On Vehicle Arrival Time Prediction Algorithm Based On Urban Public Transportation Trajectory Data

With the acceleration of urbanization in China,urban traffic congestion has become an important issue restricting urban development.In recent years,people from all walks of life in the community have generally realized that it is necessary to alleviate urban congestion by developing public transportation.Bus travel has the characteristics of convenience,flexibility and low cost,and carries the travel needs of the general public.However,due to the interference of accidental and sudden factors during the operation of the ground bus,there is great uncertainty.For the public,the uncertainty of bus travel will affect the comfort of bus travel and reduce riding Willingness to travel by bus.Studies have shown that bus arrival time is one of the public's most concerned travel issues.With the popularization of automatic vehicle positioning systems,the prediction of bus arrival time has become an important topic for many scholars to invest in research.The operation process of the bus is divided into the stop stage and the inter-station driving stage.In order to improve the accuracy of the prediction,models are proposed to predict the time of these two stages.Aiming at the problem of single data source used in the existing road average speed calculation method,a road average speed algorithm based on data merging is proposed.The algorithm integrates the global factors and local factors of the vehicle in the process of traveling,and data fusion of the average speed of the multi-vehicle section and the instantaneous speed of the multi-vehicle to obtain the integrated average speed of the section.Experimental results show that,compared with the speed integral algorithm,the calculation results of the algorithm are reduced by 7.29%,9.53%,and 13.8%in MRE,MARE,and max(ARE),respectively.Aiming at the existing traffic flow velocity prediction mainly providing point prediction,this paper proposes a combined model to provide interval prediction.The model extracts the linearity of real-time data through time series and generates a residual sequence.Fuzzy information granulation is used to generate fuzzy information granular of residual sequences.Fuzzy support vector machine is used to predict the fuzzy information granular.Finally,the prediction result of the fuzzy information granular is combined with the prediction value of the time series to calculatel the interval prediction of the traffic flow velocity.The experimental results show that the interval prediction calculated by the combined TFFS model can better describe the changing range of traffic flow velocity.In this paper,the GA-BP model is used to predict the stop time of the station,and the factors that affect the stop time of the bus station are used as variables into the model to predict the stop time of the station.Finally,the prediction results of the bus arrival time are obtained by combining the prediction results of the stop time of the station and the travel time between the stations.Experiments show that the method proposed in this paper has good practical value in short-range prediction.