Characteristic Analyses And Estimations Of Travel Times In Urban Roads

Travel time is one of the most crucial and visual indices to evaluate traffic performance,which has been investigated for decades of years.In urban roads,travel time is impacted by traffic demands,traffic supplies(including traffic signal controls),and traffic environment.Among these factors,traffic signal control is the most important,which determines the interrupted characteristic of urban traffic flows.In this paper,the influence of signal control is considered in the process of analyzing and modeling of urban travel times.The field data collected in four arterials in Guiyang,China is first used to explore the data characteristics in the frequency distribution histograms and the overlay charts of travel times.In the frequency distribution histograms,the number of peaks is influenced by the time windows and the inclination directions of data strips.Besides,the majority of travel times' statistical distributions are positively skewed and leptokurtic.In the overlay charts of travel times,the data samples present some special plane distribution characteristics,such as intermittent occurrences,clustering features,stratified features,and striped features.The striped features are related to the cycle lengths of upstream and downstream intersections,which won't be influenced by the calculation ranges of travel times.For further understanding the plane distribution characteristics,the differences between vehicles'upstream passing times,as well as the differences between their travel times,are calculated under the assumptions of first-in-and-first-out,saturated releasing,and no-spillover.Some special features are observed during the process of analyses,including clustering features,linear change,stratified features,striped features,and circulation features.These plane distribution characteristics can be quantified by cycle lengths,effective green times,and saturated headways.The changes of cycle lengths and effective green times lead to the changes of the offsets and the number of vehicles on the roads,thus influencing the plane distribution characteristicsWith the consideration of travel times' clustering features,a travel time estimation method based on the vehicle group is proposed to select representative values from travel time observations obtained from point-to-point detectors.The travel time estimation problem is transformed into a clustering problem by recognizing vehicle groups and group centers.Those recognized group centers with dense densities,defined as "grouped travel times",are used to represent the trarvel times of traffic flows,which can also reflect the influences of signals and capture the multistate features of travel timesAnother travel time estimation method is proposed to estimate a single vehicle's travel time.Under the assumptions of first-in-first-out and point queue,the vehicle's movement process on a single-lane can be described by "block function" and "green or red function".These two functions are used to determine whether a vehicle is blocked by the front vehicle or the red light,which helps to divide vehicles into four categories.Then,the travel times of vehicles in each category can be calculated based on vehicles' releasing features at signalized intersections.In this paper,a series of analyses have been done to investigate the data features in frequency distribution histograms and overlay charts of travel times.Besides,two travel time estimation methods have been proposed to estimate the travel times of urban traffic flows and single vehicles.The findings of present study provide a solid foundation for Intelligent Transportation Systems and traffic control systems,which also help us to understand urban traffic flows more deeply.