Development of ground motion prediction equation for Agean Region in Turkey based on gene expression programming

Site-specific seismic hazard analysis is one of the important processes in determining the ground motion parameters required for earthquake-resistant structure design. The selection of ground motion prediction equations to be used in seismic hazard analysis studies is one of the most influential factors on analyses success. Within the scope of this study, ground motion prediction equation models were created based on strong ground motion database recorded in the Aegean Region, Turkey. The database obtained through the AFAD (Disaster and Emergency Management Presidency) system consists of 463 acceleration records of 148 earthquakes. The database used to create the ground motion prediction model has a magnitude range of 4.5≤Mw≤6.5 and a source-site distance of 0≤Repi≤212 km. In recent years, the use of artificial intelligence-based machine learning tools has increased in the field of geotechnical engineering, especially to model the complex behaviour of soils. In this thesis, four different ground motion prediction models based on Gene Expression Programming (GEP) were created using artificial intelligence. In two of these models, largest value of the horizontal acceleration components (PGAmax) has been considered, and in the other models, the geometric mean of the horizontal acceleration components (PGAgeo) has been used. In two of the four prediction models the effect of local soil conditions has been included with the average shear wave velocity parameter in the upper 30 m (Vs,30). Different statistical performance parameters were examined to evaluate the applicability of these models. In addition, the performance of these GEP-based models was compared with four different estimation equations developed based on regression for Turkey and its surroundings. These comparisons and analyses show that the GEP-based models created for the Aegean Region can be successfully used to predict the maximum ground acceleration. Keywords: Gene Expression Programming (GEP), Earthquake, Peak Ground Acceleration, Ground Motion Prediction Equation, Aegean Region