Research On The Prediction Of The Arrival Time Of Urban Public Transport Vehicles

Accurate prediction of the arrival time can facilitate the reasonable planning of travel time for passengers who choose to take buses,reduce the waiting time of passengers at the station and provide support for bus scheduling.The driving of public transportation vehicles will be affected by various factors such as weather conditions,air quality conditions and urban road congestion.In order to effectively predict the arrival time of public transportation vehicles,this thesis take Qingdao bus arrival time prediction as the target and gives a solution.The specific work is as follows:Considering the influence of weather conditions and air quality conditions on the arrival time of public transportation vehicles,based on the pre-processing of bus-arrivalsdepartures-station data,weather condition data and air quality condition data in Qingdao,The association rule algorithm is used to analyze and extract the relevant feature attributes that affect the time of the bus stops at the station and the travel time between stations.A bus arrival time prediction model was established.The model divides the bus arrival time prediction into two parts: the time of bus stops at the station prediction and the travel time between stations prediction.The time of bus stops at the station prediction model and the inter-station road travel time prediction model are established by the multiple regression model and the integrated algorithm model respectively.Through the experimental analysis,the best predictive performance of the GBRT-based bus arrival stop time prediction model and the XGBoost-based inter-station road travel time prediction model are finally obtained.In order to further improve the prediction accuracy of the bus arrival time prediction model,considering the distribution difference of data in different periods,establish the bus arrival time prediction model for the morning peak,the evening peak,and the non-morning and evening peak respectively.The experimental results show that the model is compared with the overall training model and the bus arrival time prediction model established by dividing the morning peak,the evening peak,and the non-morning and evening peaks respectively has better prediction accuracy.