Development of a hierarchical data-driven pavement deterioration prediction model with an adaptive approach for improving resilience, pedestrian safety, and maintenance optimization in urban design.

Alireza Zareyi¹

1) Master s degree in Urban Design, Tabriz Metropolitan Municipality, Iran

IOI Code : AACZ-ABGDBC
Publication : The 5th International Conference on Seismic Retrofitting(5icst.ir)
Abstract :
    Pavement deterioration in urban thoroughfares, especially at intersections as focal points of traffic-induced stresses (acceleration, braking, torsion), is one of the main factors reducing the resilience of the transportation network, increasing pedestrian safety risks, and imposing maintenance costs on urban management. Traditional analytical models, due to simplifying traffic behavior and ignoring local deterioration patterns, lack sufficient accuracy for prediction at intersection and network scales. This research aims to develop a hierarchical data-driven model. The proposed model operates on two levels: the first level provides local deterioration prediction at intersection nodes using machine learning algorithms (e.g., Random Forest) based on real-time traffic, geometric, and climatic data. The second level generalizes the results to the entire road network using an adaptive approach (integrating analytical and data-driven methods) to optimize the timing and location of maintenance operations. The model output provides risk zoning maps of deterioration and prioritization of interventions, which can be directly used in the geometric design of thoroughfares (slope, radius, pavement) and improving pedestrian safety. This approach represents a step toward resilient, data-driven, and cost-efficient urban design.

Keywords : hierarchical data driven model pavement deterioration prediction machine learning resilient urban design