Safety Analysis Of Signal-Coordinated Arterials Considering The Heterogeneous Correlations

The implementation of signal coordination on the arterials has drawn an increasing attention in the recent decades.Different from the isolated control strategies,signal coordination couples the phases of multiple intersections simultaneously,aiming to improve the operational efficiency for the passing vehicles.Specifically,the coordination of signalized intersections creates a “green wave,” which allows the vehicular platoon to consecutively move along the arterials with reduced stops.The majority of the previous studies verified that the signal coordination had a number of advantages,such as reducing the travel time and delay and decreasing the vehicular emissions.Previous studies mainly focus on assessing the efficiency of signal coordination,while limited literature systematically aims to evaluate the safety performance.Meanwhile,the safety impact of coordinated signalization remains controversial in the limited research available.To avoid the potential crash risks and reveal the specific safety status,thus it deserves an in-depth exploration of arterials under the signal coordination.The accurate evaluation serves to provide the scientific basis for the safety improvement of the signal-coordinated arterials.The evaluation model of crash frequency is a useful method to unveil the safety status of arterials with the coordinated signalization.During the modeling process,the unobserved heterogeneity among traffic crashes is generally ascertained as a critical issue that should be non-negligible.The heterogeneity is formed in the crash data source with the inherent differentiation that cannot be witnessed directly from the observations.Ignoring the heterogeneous associations may cause the inconsistent parameter approximations and erroneous accident predictions in the safety analysis.Given the characteristics of traffic flow formed on arterials the under signal coordination,the exploration of heterogeneous correlations should focus on the microscopic,mesoscopic,and macroscopic levels during the formulation procedure of safety assessment models.Therefore,the thesis proposes the comprehensive theory of crash frequency analysis by incorporating the multiple heterogeneities into the same framework.The multidimensional study of heterogeneity serves to achieve the systematic safety evaluation of signal-coordianted arterials.The paper selects six signal-coordinated arterials with the adaptive traffic coordination system in Ann Arbor,Michigan(USA).Nine-year(2006-2014)crash counts are extracted from the database provided by Michigan Department of Transportation.At the microscopic level(i.e.,on the roadway segments and at the signalized intersections),the study adopts the multivariate Poisson log-normal(MPLN)model to reveal the mutivarite heterogeneity between injury and property damage only(PDO)crashes along the coordinated arterials.In addition,spatial correlations of conditional auto-regression(CAR)and multivariate conditional auto-regression(MCAR)are added to the MPLN model to calibrate the regressive results.Based on the surveyed arterials in Ann Arbor,Michigan,the case study shows that 1)the MPLN-MCAR model predicts the results more accurately compared to the MPLN-CAR and MPLN models;and 2)there is a strongly correlated heterogeneity between injury and PDO crashes.At the mesoscopic level(i.e.,in the traffic analysis zones),the study aims to evaluate the structural heterogeneity on arterials with coordinated traffic signals by applying a joint negative binomial conditional auto-regression crash sum(JNBCS-CAR)model.The performance of JNBCS-CAR is compared to those of traditional five ones(i.e.,Poisson log-normal(PLN),negative binominal crash sum(NBCS),Poisson log-normal conditional auto-regression(PLN-CAR),negative binominal conditional auto-regression crash sum(NBCS-CAR),and hierarchical Poisson log-normal conditional auto-regression(HPLN-CAR)).Nine years of data are selected from the City of Ann Arbor(Michigan).Results indicate that 1)the JNBCS-CAR model outperforms other models in terms of the goodness-of-fit;and 2)signalized intersections appear to have significant spatially heterogeneous effects on roadway segments under the scenario of signal coordination,while roadway segments do not have spatially heterogeneous effects on signalized intersections.At the macroscopic level(i.e.,along the coordinated arterials),the paper compares three types of spatiotemporal heterogeneity with the PLN model on the signal-coordinated arterials,including the fixed spatial and linear temporal variation,the fixed spatial and linear-quadratic temporal variation,and the time-varying spatial interaction.With the use of9-year crash data from Ann Arbor,Michigan(US),the study suggests that 1)different models should be separately applied for the coordinated arterials,e.g.,the PLN model with the time-varying spatial interaction best fits the data of roadway segments,and the PLN model with the fixed spatial and linear temporal variation excels at signalized intersections;and 2)the temporal variation and spatial variation are correlated on the roadway segments,but not at the signalized intersections.The findings emphasize the importance of conducting the safety analysis regularly and dynamically monitoring the driving behaviors especially along the roadway segments.At the multi-dimension level,the thesis formulates the model by incorporating multiple heterogeneities from the microscopic,mesoscopic,and macroscopic aspects simutaneously.The model suffices to provide the comprehensive safety assessment of arterials with the coordinated signalization.Modeling results imply that 1)there exist multiple heterogeneities(i.e.,multivariate,structural,and spatiotemporal heterogeneities)along the coordinated arterials;and 2)multivariate heterogeneity has the strongest impact on the safety evaluation results,followed by the spatiotemporal and structural heterogeneities;3)the variation of spatial heterogeneity is unrelated to the temporal changing among the traffic analysis zones.In an attempt to alleviate the correlated crash risks,the study demonstrates the necessity of setting the sub-control traffic analysis zones in real-time based on the criterion of correlative degree.With condering the heterogeneous correlations,the study employs the appropriate models to identify the significant contributors,thus achieving the accurate safety evaluation of arterials with the signal coordination.The findings serve to provide effective countermeasures for safety planning,design,and management of the signal-coordinated arterials.Meanwhile,the crash prediction framework of multiple heterogeneities can be applied into the safety analysis of other urban roads.