Support Vector Machine Method for Developing Ground Motion Models for Earthquakes in Western Part of China

This article develops ground-motion models (GMMs) by using a support vector regression (SVR) for western part of China. These SVR GMMs constitute mathematical equations that can predict the peak ground acceleration (PGA) and response spectral at different periods. Given earthquake magnitude (Mw), hypocentral distance (Rhypo) shear-wave velocity averaged over top 30 m soil (Vs30), predicted values be acquired. The results demonstrate proposed are reasonable produce accurate estimations with good generalization capability motions exhibiting range seismic characteristics similar to those considered in database.