On the prediction of shear wave velocity at local site of strong ground motion stations: an application using artificial intelligence

Since the determination from experimental tests are expensive and time consuming, the site conditions in strong ground motion equations are mostly expressed by geologically qualitative descriptions of soils at the recording stations. The analytical solution for the site description has not been sufficiently studied due to highly nonlinear behavior of soil. Advances in field of artificial intelligence (AI) offer new insights to solve the problems in the most complex systems utilizing different algorithms and models. This paper primarily aims to predict average shear wave velocity (V-S30) as a soil property at the earthquake recording stations by applying AI methods, which are composed of artificial neural network (ANN) and genetic expression programming (GEP). The application is performed for the 60-accelerograph station sites located in California, USA. The predictor variables of V-S30 in AI models, which are properly organized from strong ground motion data, are magnitude, site-to-source distance, peak ground acceleration and spectral accelerations at different site periods. values as output variable are collected from the surface wave testings conducted in the sites. The results indicates that for the considered highly nonlinear problem in this paper, the developed ANN and GEP models perform good predictions in terms of error and correlation. It can be concluded that the AI methods are relatively promising for prediction of V-S30. The findings from this paper can be helpful to improve the site descriptions at the current database of the study region.