Probabilistic seismic response transformation factors between SDOF and MDOF systems using artificial neural networks

Abstract

An approach to obtain with acceptable accuracy probabilistic response transformation factors by training an artificial neural network (ANN) model is presented. The transformation factors are defined as the ratio of the seismic response of multi-degree-of-freedom structures and their equivalent single-degree-of-freedom systems, associated with a given annual exceedance rate. The approach is used for predicting the seismic response of steel framed buildings. Equations useful to obtain probabilistic response transformation factors for maximum ductility and inter-story drift, as functions of their mean annual rate of exceedance, and of the fundamental vibration period of the structure, are proposed. It is shown that artificial neural networks are a useful tool for reliability-based seismic design procedures of framed buildings and for the improvement toward the next generation of earthquake design methodologies based on structural reliability.