Urban Traffic Congestion Modeling Based On Taxi GPS Data

While urbanization and the increasing vehicles facilitated modern transportation,traffic congestion has become one of the major "urban diseases",mainly indicated by the worsening degree of congestion and extended overcrowding periods.Therefore,it is important to control the traffic status of road in the traffic management.Taxi with GPS is an efficient measurement to detect traffic condition,and the research on urban transport problems using this measurement has been given more attention in recent years.This thesis focuses on the traffic congestion prediction modeling using taxi GPS data.The contents include:(1)Preprocessing of taxi GPS data.The research analyzes and preprocessed the taxi GPS data in the original txt format,including analysis of data format and field meanings,error analysis and error handling methods,road network matching principles,filtering out error data,establishing data processing principles and methods,the conversion of data format which make it available for the next analysis,sorting the data in chronological order and so on.(2)Constructing traffic parameter prediction model based on spatial correlation.There are three different traffic parameter prediction model:ARIMA model,RBF neutral network model and Combination model.When using ARIMA model to predict traffic flow parameters,previous prediction results are analyzed continuously,while future prediction values are adjusted according to their prediction errors.RBF neural network prediction model can handle more input data and more complicated situation.The optimal weight distribution method is used to predict traffic flow parameters for the combination of the two models.(3)Constructing congestion identification model based on speed.The research studies the traffic congestion definition,then selects speed as the main traffic parameter for prediction and evaluation.A congestion identification model based on speed is constructed based on the definition of the evaluation criteria.