The applications of artificial intelligence for magnitude types conversion and peak ground acceleration (PGA) prediction of earthquakes

Turkey has a high seismic hazard due to the inclusion of active tectonic structures such as the North Anatolian Fault Zone, the East Anatolian Fault Zone and the Aegean System. Therefore, seismic hazard assessment must be reliable and have high accuracy. In seismology and hazard assessment, magnitude type conversion equations and peak ground acceleration (PGA) values are used for a homogeneous earthquake magnitude type and for determining the effects of earthquakes on engineering structures. The goal of this thesis, Artificial Neural Networks (ANN) models, regression models and equations have been developed to calculate PGA values of earthquakes and conversion of earthquake magnitude types by using Artificial Intelligence (AI) techniques such as ANN, Meta-heuristic algorithms and Machine Learning (ML) algorithms. For magnitude type conversion 8 equations and 3 regression models were created by using regression methods and ML algorithms by values and magnitude (M ≥ 4.0) types of earthquakes (1900-2020) occurred in Turkey and its surroundings. For PGA prediction, ANN models and 3 regression models were created using ANN with Meta-heuristic algorithms and ML algorithms according to 6 different data forms with 2 data sets consisting of parameters of earthquakes (M ≥ 3.0 and Mw ≥ 5.0). All models were compared with the magnitude type conversion and PGA prediction equations selected from the literature according to the test data. Regression models formed for the magnitude type conversion gave close results to the real values. The formed ANN models and regression models calculated close values to the measured PGA values from the compared PGA prediction equations. In particular, close PGA values to the measured PGA values predicted by AI techniques will increase the reliability of seismic hazard assessments. In addition, it has been observed that AI techniques can be used effectively in seismology and engineering seismology.