Influence of simulated traffic on foundation pit deformation via machine vision technology.

Abstract

This study introduces a novel application of non-contact computer vision technology to simulate and analyse the impact of traffic loads on deformation in soft soil foundation pits, using vehicle flow data captured by roadside cameras. The research employs the YOLO (You Only Look Once) detector coupled with a multi-object tracking algorithm for vehicle classification and tracking. Following statistical analysis of traffic parameters, Monte Carlo sampling is utilised to simulate actual traffic loads. These loads are then input into a PLAXIS 3D finite element model to examine their effects on support structures and soil deformation. The comparison of measured and simulated data confirms the accuracy of the traffic load simulation model. Results demonstrate that vehicular loading can increase pit deformation by up to 48.7%, with significant impacts extending over 101.3 m along the site, significantly influencing engineering planning and construction management. This study provides a scientific basis for safety assessments in urban infrastructure projects, demonstrating the potential of non-contact machine vision technology for analysing the impact of traffic loads on the safety of adjacent excavation projects.