A Machine Learning-Based Framework for Predicting Earthquake-Induced Building Debris Volume and Estimating Heavy Equipment Requirements
A Machine Learning-Based Framework for Predicting Earthquake-Induced Building Debris Volume and Estimating Heavy Equipment Requirements
Vahideh Nemati1
1) M.Sc. in Civil Engineering – Earthquake Engineering, Tabriz Metropolitan Municipality, Tabriz, Iran Email:
IOI Code
:
ABKD-ABHFFK
Publication :
Fourth International Conference on applied researches in civil engineering, architecture and urban planning (icarcau.com)
Abstract :
Following destructive earthquakes, the massive volume of debris generated by building collapse represents one of the most critical obstacles to emergency response, rescue operations, and urban recovery. Debris-induced road blockages can severely restrict emergency teams’ access to affected areas and delay the critical golden period for rescuing injured individuals. This study investigates the application of machine learning techniques to predict the volume and spatial distribution patterns of debris resulting from seismic building damage and to estimate heavy equipment requirements for road clearance operations. The research methodology is based on a systematic review of relevant studies and an analysis of advanced modeling approaches in this field. The findings indicate that machine learning algorithms, including Gradient Boosted Decision Trees (GBDT) and Random Forest (RF), can predict debris distribution patterns with high accuracy, achieving coefficients of determination (R²) exceeding 0.93, and can generate heat maps representing the spatial distribution of earthquake-induced debris. Furthermore, machine learning-based surrogate models can reduce computational time by more than 90% compared with conventional finite element methods. The results demonstrate that integrating physics- and engineering-based models with machine learning algorithms provides an efficient approach for estimating logistical and heavy equipment requirements in post-earthquake disaster management. By proposing a comprehensive framework for debris prediction and clearance resource estimation, this study contributes to enhancing urban seismic resilience and improving post-earthquake emergency response planning.
Keywords : seismic debris prediction machine learning road clearance disaster management heavy equipment urban resilience