Study on ring-road incident duration based on latent class accelerated hazard model.

Area of Science:

• Transportation Engineering
• Traffic Management Systems
• Statistical Modeling

Background:

• Accurate freeway incident duration estimation is crucial for effective emergency response and preventing secondary incidents.
• Understanding the factors influencing incident duration is key to optimizing traffic management strategies on freeways.

Purpose of the Study:

• To investigate the mechanisms through which key factors influence freeway incident duration.
• To develop and apply a latent class accelerated hazard model incorporating heterogeneity to analyze incident duration.
• To identify significant variables and their impact on freeway incident duration, particularly on ring roads.

Main Methods:

• A latent class accelerated hazard model was developed and applied to freeway incident data.
• The Log-logistic distribution was selected over Weibull and Log-normal distributions for superior performance.
• Statistical analysis identified 11 significant variables influencing incident duration across two latent classes.

Main Results:

• Two distinct latent classes were identified with probabilities of 0.53 (Class 1) and 0.47 (Class 2).
• Specific lane closures and driver reporting methods were found to be significant factors influencing incident duration.
• The model captured potential heterogeneity and its influencing mechanisms, with most incidents falling into Class 1 (0-60 minutes duration).

Conclusions:

• The study provides insights into reducing incident duration on Chinese freeways, particularly ring roads.
• Findings offer theoretical support for developing effective freeway incident management and treatment policies.
• The model's ability to capture heterogeneity and specific influencing factors enhances understanding of incident dynamics.