Predictive Model of Artificial Neural Network for Earthquake Influence Analysis

Earthquake affecting coefficient is a main parameter of earthquake response spectra, which is the foundation of seismic microzonation and design of earthquake resistant structures. Because the distribution of the maximum of earthquake affecting coefficient is controlled by both basement rock and site condition, the relationship between the maximum of earthquake affecting coefficient and influencing factors is complicated. In order to design earthquake response spectra subtly, the predictive model is constructed on the basis of artificial neural network (ANN), which makes the distribution of the maximum of earthquake affecting coefficient become a spatial variable. As an example application in Tangshan City, the distribution of the maximum of earthquake affecting coefficient is calculated precisely. The calculating results are analyzed and some advice is proposed.