APPLICATION OF ARTIFICIAL NEURAL NETWORKS FOR GENERATING STRONG GROUND MOTION PARAMETERS

Abstract

Artificial Neural Networks (ANNs) are efficient computing models which have shown their strengths in solving complex problems in many fields. They have the versatility to approximate a wide range of complex functional relationships between sets of input and output data. The purpose of this study is to predict strong .ground motion parameters that are of primary significance in earthquake engineering using ANN. In this study, sets of Multilayer Perceptron neural network model are trained to predict the normalized response spectra, Zero Period Acceleration (ZPA), and duration of strong motion using Japanese earthquake records and site characteristics. The database used in the study is taken from Kyoshin Net (K-NET) database of Japan. NeuroSolutions (neural network simulator) software has been used to model ANN and the standard back-propagation supervised training scheme is used to train all networks. ANN has been used to solve the problem of predicting strong motion parameters using records of Japanese earthquakes of magnitude more than 5.0 and hypocentral distance less than 50 km. In this study, 1850 horizontal components of time histories have been used. Basic information such as magnitude, hypocentral distance, and average values of Standard Penetration Test (SPT) blow count; primary wave velocity; shear wave velocity, and density of soil have been used as six input variables to train the neural network. Since at most locations in world, only average shear wave velocity is used for site characterization, therefore, an attempt has also been made to train the neural network with magnitude, hypocentral distance, and average shear wave velocity as three input variables using Japanese records.