IDENTIFICATION OF DAMAGE IN STEEL BEAM BY NATURAL FREQUENCY USING XGB MODEL

Abstract

Beams have played a significant role in engineering applications and they have been commonly used for modelling civil problems. Different models and methods have been developed to identify the damage to the beams. In this article, the extreme gradient boosting (XGB) model was developed to predict the location, width and depth of the saw-cut of steel beams by the change of natural frequencies. The natural frequencies of a steel beam in different scenarios were identified by the finite element method (FEM). The criterions to evaluate the accuracy of the models were the R squared (RSQ) and the mean square error (MSE). The result indicated that combining the FEM method with XGB would hold significant potential for applications in structural health monitoring.