A physics-based approach for damage identification in plates based on guided wave scattering predictions and inverse Bayesian inference

Ultrasonic guided waves are one technique suitable for damage detection in thin plate like structures. One can characterize the plate damage by analyzing scattered fields generated by wave damage interaction. A semi-analytical forward model is presented to perform rapid computations on wave scattering which provides key elements for damage characterization. A fast Lamb-wave based physics-informed damage identification Bayesian framework in 2D structures is proposed to identify the geometry of a partly through-thickness circular hole using wave scattering properties. A numerical case study is used to verify the proposed damage identification scheme. The scattering coefficients are reconstructed, which has a good correlation with true values. The framework used in this work is about 1,676,470 times more efficient than the one based on explicit finite element model. The main limitation of the proposed framework is that the identified damage must be spherically symmetric, but such defect forms are very representative in metal structures.