Distribution of Peak Ground Velocity in Miyagi Prefecture Estimated from One Accelerogram Using Artificial Neural Network

摘要

In this study some of the advantages of ANN, using experiences, learning capability and constituting relation past and future or inputs and outputs are employed to get the prediction of N-S component of PGV distribution in Miyagi prefecture based on one accelogram records using Artificial Neural Network method is achieved. The testing result values maybe not exactly fit the experienced data as training sets however the distribution maps show good agreement with real distributions. It is convenient to use large amount of data in ANN but also it is obvious that as more recorded accelograms or earthquakes become available they can be used to retrain the ANN to get better results.