No abstract

In this paper, a damage quantification strategy relying on post-processing of Lamb wave based damage localization results is presented. This method is able to predict the upcoming sizes of a delamination after a training step. Inputs of the proposed method are localization index maps produced by damage localization algorithms and representing the presence likelihood of a damage over the structure under study. The area covered by a high localization index around the estimated damage location are then extracted from these spatial probability maps. A data-driven model representing the mathematical relationship between this quantification feature and the actual size of the damage is finally inferred and used to predict future damage size. The proposed method is successfully validated on experimental data coming from CFRP plate samples equipped with piezoelectric transducers. Delaminations induced by fatigue testing and laser shock are studied. The sensitivity of the method to input frequency and damage localization algorithms parameters is assessed and a method to automatically select its own parameters is proposed. Furthermore, it is demonstrated that a model can be confidently learned on a given CFRP plate sample and transferred to predict damage size on another similar CFRP plate sample.

No abstract