Generating Artificial Earthquake Ground Motion Based On A Time-domain Stochastic Ground Motion Model

A time-domain stochastic ground motion model(SGMM)is an important research topic in earthquake engineering.Developing a time-domain SGMM based on recorded ground motions and generating synthetic ground motions for a given earthquake scenario are the bases of seismic analysis.In this dissertation,a time-domain SGMM is developed by combining a filtering white noise model and a seismic attenuation relationship.The developed model can generate synthetic ground motions for a given earthquake scenario,which has a great significance for seismic analysis.The main contents of this dissertation can be summarized as follows:(1)Literature review is conducted on the research of stochastic ground motion models and seismic attenuation relationships.Different types of models are compared and discussed.(2)The filtering white noise model is introduced.Firstly,the deviations of the formula are carried out.Secondly,a semi-statistical-physical-based method is used to identify the parameters of the model.Finally,ground motions are simulated by the proposed model.The results show that the simulated ground motions can properly capture the characteristics of the measured records.(3)Research on the seismic attenuation relationship is conducted based on Bayesian inference.The derivations for Bayesian Regression,Bayesian Network and Gaussian Process Regression are presented.Three seismic attenuation relationship models are constructed accordingly,and crossing validation is adopted to select the optimal model.(4)Combining the filtering white noise model and the optimal seismic attenuation model,this study develops a time-domain SGMM.This developed model is capable of generating ground motions with giving an earthquake scenario,which can be utilized in structural seismic analysis and design.