Importance Analysis of Structural Seismic Demand Based on Support Vector Machine

Abstract

Seismic demand analysis of structures plays an important role in the structural seismic calculation; however, studies on the importance analysis of seismic demand are limited. A new method based on a support vector machine (SVM) is proposed to analyze the importance of structural seismic demand and study the influence of random variables on structural seismic demand in this study, where the linear kernel function, Gauss kernel function, and polynomial kernel function are used in SVM. The time history analysis of the steel-reinforced concrete (SRC) frame structures has been carried out by the finite element software OpenSees under the action of different seismic records. Four kinds of seismic demand of the SRC frame structure are analyzed in this study, which are top displacement, maximum floor acceleration, base shear, and maximum interstory drift angle, respectively. Importance indexes of the four kinds of structural seismic demand are in good agreement with those of the Monte Carlo (MC) numerical simulation method and Tornado graphic method, which verify the accuracy of the proposed method. Moreover, the sample size of the proposed method is greatly smaller than that of the MC method. Therefore, the computation efficiency has been improved significantly by the proposed method.