Probabilistic Seismic Hazard Analysis of a Back Propagation Neural Network Predicting the Peak Ground Acceleration

Probabilistic seismic hazard analysis (PSHA) has been recognized as a reasonable method for quantifying threats. Traditionally, this ignores the effect of focal depth, in which ground motion prediction equations (GMPEs) are applied to estimate probability distribution associated with possible levels induced by site earthquakes, but it is limited unclear geological conditions, makes difficult provide uniform equation, and these cannot express non-linear relationship under conditions. Hence, paper proposed consider depth PSHA example California used back propagation neural network (BPNN) predict peak acceleration (PGA) instead GMPEs. Firstly, measured PGA unknown data applicable were collected separately. Secondly, supplemented applying BPNN based on data. Lastly, full-probability considering was completed compared current zoning results. The results showed that using can ensure computational accuracy universality, making more suitable regions structures providing possibility adding other parameters be considered influence PSHA.