OVERVIEWING THE APPLICATION OF SURROGATE MODELS IN THE ANALYSIS OF RELIABILITY AND RISK OF STRUCTURES AFFECTED BY RANDOM FACTORS

Abstract

Abstract: The study presents
an overview of the role of surrogate
models in reliability and risk analysis
of structures under seismic, wind,
and multi-hazard effects. The
surrogate models significantly reduce
computational costs compared to highfidelity simulations while maintaining
acceptable accuracy. Representative
approaches such as Kriging, Gaussian
Process, Artificial Neural Networks (ANN),
and Polynomial Chaos Expansion (PCE)
are analyzed in detail. Furthermore,
current challenges such as training
data reliability, nonlinear behavior,
and multidimensional uncertainties are
highlighted. Finally, the study suggests
an orientation of integrating surrogate
models with artificial intelligence and
high-performance computing to support
sustainable infrastructure development
in the future.