Central AsianResearch Index

FLOOD RISK PREDICTION FOR BRIDGE INFRASTRUCTURES WITH MACHINE LEARNING

Flood prediction is required for infrastructure protection and disaster management. Tra- ditional physics based models are limited in representing complex nonlinear relationships among en- vironmental and infrastructure factors. Machine learning, i.e., Long Short Term Memory, Random Forest, and Support Vector Machine models are evaluated for the prediction of exposure, criticality, andoverallrisk. TheresultsshowthatRandomForestprovidessuperiorperformanceforexposureand criticality assessment, while Long Short Term Memory achieves the best performance for overall risk prediction. The findings indicate that different machine learning methods are suited to different flood risk components. A comprehensive approach to flood risk assessment is therefore supported. Valu- able information is provided for infrastructure planning, risk reduction, and decision making. Further improvement is expected through the inclusion of additional data sources and combined modeling methods.

Сведения

Опубликовано
2026 · Т. 14, № 2, С. 67-77
Язык
Английский
Цитирований
1
ISSN
2306-6172 · 2308-9822
Издатель
L. N. Gumilyov Eurasian National University
Аффилиации
James Watt School of Engineering, University of Glasgow
Obuda University