A comparison of different surrogate models for delamination detection in composite laminates using experimental modal analysis

Abstract

This paper deals with the comparison of different surrogate models such as kriging surrogate model, support vector regression, radial basis functions and polynomial response surfaces for locating the damage and estimate its size for delamination detection in composite laminates using frequency shifts derived from experimental modal analysis. The strategy involved a forward problem and inversion using surrogate models. The data for the inversion was generated by Finite element analysis and experimental Modal analysis and was validated using the modal assurance criterion. It was found that the predictions made using kriging surrogate model and Radial basis functions had the lowest errors when compared with the other two models namely the polynomial response surfaces and the support vector regression.