Research On Freeway Travel Time Prediction And Travel Time Reliability Based On Toll Data

Freeway is an important carrier of medium and long distance road transportation.It is an important channel for the rapid delivery of people and goods.The operating characteristics of freeway are complex.The accurate judgement of various operating characteristics can better improve the operating efficiency of highway systems.The enormous toll data of freeway contains a mass of valuable traffic information.Trough the researching and analyzing of these data in different dimensionality can obtain various of freeway operating characteristics.So that it can provide valuable travel information for travelers,and also it is helpful for traffic managers to better manage the entire road network system.The paper mainly uses the average speed data of Shandong freeway for freeway travel time prediction and travel time reliability evaluation.The paper first sorts out and summarizes the studies on travel time prediction and reliability evaluation that had been made at home and abroad.Travel time prediction research also contains the traffic information data preprocessing methods.The current travel time reliability research contains the definition of travel time,and the summary of measurement indicators and measurement methods.Secondly,the performance characteristics of the expressway were analyzed from different perspectives using the existing data collections which includes the traffic flow data sets and average speed data sets.Then the paper forecasts the travel time of the freeway and evaluates the travel time reliability based on the average speed dataset.Based on the idea of data mining in travel time prediction,k-means clustering algorithm based on improved artificial bee colony algorithm is used to cluster the travel time at different time periods.A weighted estimation method is used to predict the travel time of the cluster centers,and a number of road segments are randomly selected to use the average speed dataset for the prediction method verification.The results show that the prediction accuracy of each period is high.In the travel time reliability evaluation research,the paper uses the idea of probabilistic to solve travel time reliability.Based on the multi-peak characteristics of the travel time distribution,the Gaussian mixture distribution is used to fit the probability density of travel time.Using EM algorithm for parameter estimation to obtain the parameters of Gaussian mixture distribution before fitting.The goodness-of-fit test was performed after the fitting.The goodness-of-fit evaluation parameters showed that the fitting effect was good.Finally,summarize the research results of the paper and put forward the future research prospects.