Delamination Localization in the Composite Thin Plates Using Ensemble Learning: Bagging and Boosting Techniques

Abstract

Localization of the delamination is an essential task that is conducted via various approaches, which may require time, experts, and cost. Various intelligent nondestructive techniques are utilized to reduce time consumption, the need for expertise, and expenditures. Yet, developing an accurate, robust, and low-cost intelligent delamination identi cation technique becomes a challenging task due to the anisotropy and the variation in the ber orientation of the composites. Based on those issues, it is aimed to develop an e ective intelligent model to localize delaminations in composite plates. This study measures the performance of the Bagging and Boosting techniques on delamination localization in thin composite plates. To validate the e ectiveness of the proposed approaches; cross-ply, angle-ply, and quasi-isotropic composite plates having 2400 di erent delamination cases are considered. The bagging and boosting models are trained with the vibrational characteristics of the healthy and delaminated composite structures. The free vibration analysis is conducted for those structures to obtain the rst ve natural frequencies and the corresponding mode shapes. For this purpose, classical plate theory is employed by using nite element analysis. It is concluded that bagging and boosting techniques are robust, precise, and accurate in localizing delamination.