Building a Two-Dimensional Two-Layers Linear Site Response Solver with Machine Learning Model

Abstract

A multidimensional site response analysis is conducted to evaluate the site amplification of ground motions in surface ground. The tools for this analysis have been well developed, and these analyses can be one of the options for seismic motion evaluation in structural design. In order to further promote this situation, a two-dimensional two-layer linear site response solver was constructed using a machine learning model. It is confirmed that the solver can be used to easily evaluate the ground amplification at the ground surface position for any two-dimensional two-layer soil structure. Also, by using the solvers, it may be possible to elucidate phenomena related to the site amplification, perform uncertainty analysis in reliability design, and perform inverse analysis of soil structures more efficiently.