The measurement of travel time reliability is of increasing importance as agencies seek to understand better the performance of their facilities and communicate trip expectations to the traveling public. Recent research suggests that modeling historical travel times with multistate distributions can add value to the process of monitoring reliability by linking travel times with the underlying traffic state. Although that research demonstrated the feasibility of breaking down travel time distributions into different operational states, the researchers did not evaluate the causal factors of the states. Models that link the travel time states of agencies' facilities with the factors that cause the time states, such as incidents or special events, allow agencies to predict travel times more accurately when these events occur in real time and enable agencies to develop targeted projects for the improvement of reliability over the long term. This paper extends existing reliability modeling in three ways. First, where past research has used bimodal distributions to model the two primary operating regimes (free flow and congested), this paper presents a methodology for determining the optimal number of states for modeling data to ensure that the distinct effects of nonrecurrent congestion events are captured. Second, the paper presents a process that is easily implemented for determining when and how the different sources of nonrecurrent congestion influence the travel time state and its variability. Last, with the development of multistate models for aggregate travel times, the paper proves that a methodology originally developed to model individual vehicle trips can be applied to a broader array of data sets.
[1]
S. Chin.
Temporary Losses of Highway Capacity and Impacts on Performance: Phase 2
,
2004
.
[2]
Hesham Rakha,et al.
Multistate Model for Travel Time Reliability
,
2010
.
[3]
Valószínűség és véletlen számítás,et al.
Bayesian Information Criterion
,
2010
.
[4]
Chao Chen,et al.
The PeMS algorithms for accurate, real-time estimates of g-factors and speeds from single-loop detectors
,
2001,
ITSC 2001. 2001 IEEE Intelligent Transportation Systems. Proceedings (Cat. No.01TH8585).
[5]
Wenjing Pu.
Analytic Relationships between Travel Time Reliability Measures
,
2011
.
[6]
Timothy J Lomax,et al.
SELECTING TRAVEL RELIABILITY MEASURES
,
2003
.
[7]
R. Jayakrishnan,et al.
Bayesian Mixture Model for Estimating Freeway Travel Time Distributions from Small Probe Samples from Multiple Days
,
2009
.
[8]
Lily Elefteriadou,et al.
Travel Time Reliability Modeling For Florida
,
2010
.
[9]
H. J. Van Zuylen,et al.
Monitoring and Predicting Freeway Travel Time Reliability: Using Width and Skew of Day-to-Day Travel Time Distribution
,
2005
.
[10]
Adrian E. Raftery,et al.
MCLUST Version 3 for R: Normal Mixture Modeling and Model-Based Clustering †
,
2007
.
[11]
Haitham Al-Deek,et al.
Using Real-Life Dual-Loop Detector Data to Develop New Methodology for Estimating Freeway Travel Time Reliability
,
2006
.