Abstract:Safety evaluation of signalized intersections is often conducted by developing statistical and data-driven methods based on data aggregated at certain temporal and spatial levels (e.g., yearly, hourly, or per signal cycle; intersection or approach leg). However, such aggregations are subject to a major simplification that masks the underlying spatio-temporal safety risk patterns within the data aggregation levels. Consequently, high-resolution analysis such as safety risk within signal cycles and at traffic movement level cannot be performed. This study contributes to the literature by proposing a new functional data analysis (FDA) approach for a novel characterization of safety risk patterns of signalized intersections. Functional data smoothing methods that can mitigate overfitting and account for the nonnegative characteristics of safety risk are proposed to model the time series of safety risk within signal cycles at the traffic movement level. Functional analysis of variance method (FANOVA) that can compare the group level differences of functional curves is used to test differences of safety risk functions among different traffic movements. A typical signalized intersection with representative signal types and channelizations is selected as the study location and approximately 1-hour traffic video data recorded by an unmanned aerial vehicle are used to extract traffic conflicts. New movement-level safety risk patterns are characterized based on the safety risk functions that can reveal the temporal distribution of risk within signal cycles. Most of the tested traffic movements have significantly distinct functional risk patterns according to the FANOVA results while risk patterns for most of the traffic movements cannot be differentiated based on the data aggregated at the cycle and approach levels. The proposed functional approach has the potential to be used for facilitating proactive safety management, calibrating microsimulation models for safety evaluation, and optimizing signal timing while considering traffic safety at more disaggregated levels.

Safety Fundamental Diagram on Freeways

Safety Evaluation for Expressways: A Comparative Study for Macroscopic and Microscopic Indicators

Empirical Approximation for the Stochastic Fundamental Diagram of Traffic Flow on Signalized Intersection

Analytical Formulation for Explaining the Variations in Traffic States: A Fundamental Diagram Modeling Perspective with Stochastic Parameters.

Incorporating safety speed analysis into highway safety assessment

Macroscopic Travel Time Reliability Diagrams for Freeway Networks

Fundamental Diagram for Signalized Arterials: an Empirical Analysis Using High-Resolution Traffic Data

Safety Evaluation Method of Freeway Segment Based on the Fuzzy Information Distribution and Risk Assessment

Feasibility of Incorporating Reliability Analysis in Traffic Safety Investigation

A Functional Approach for Characterizing Safety Risk of Signalized Intersections at the Movement Level: an Exploratory Analysis.

On the Stochastic Fundamental Diagram for Freeway Traffic: Model Development, Analytical Properties, Validation, and Extensive Applications

Modelling the fundamental diagram of traffic flow mixed with connected vehicles based on the risk potential field

Estimating Fundamental Diagram for Multi-Modal Signalized Urban Links with Limited Probe Data

The Impact of Incidents on Macroscopic Fundamental Diagrams

A Crash Surrogate Metric Considering Traffic Flow Dynamics in a Motorway Corridor

Macroscopic Fundamental Diagram Estimation Considering Traffic Flow Condition of Road Network

A freeway vehicle early warning method based on risk map: Enhancing traffic safety through global perspective characterization of driving risk

An Empirical Analysis on the Arterial Fundamental Diagram

Modelling Highway Safety with Data from Chinese Freeways

Using Empirical Traffic Trajectory Data for Crash Risk Evaluation under Three-Phase Traffic Theory Framework.

Model on Empirically Calibrating Stochastic Traffic Flow Fundamental Diagram