A structural damage localization method for composite materials based on adaptive path screening and global intelligent search

Abstract

Accurate damage localization in carbon fiber-reinforced polymer (CFRP) laminates is an essential prerequisite for structural health monitoring (SHM). Nevertheless, Lamb wave propagation in anisotropic composite laminates is direction-dependent, dispersive, and highly susceptible to structural discontinuities; these effects often invalidate the underlying assumptions of the conventional elliptical trajectory localization method (ETLM). In practice, many actuator-sensor paths have weak interactions with damage in the sensor network. The sensor responses can be contaminated by coherent noise, boundary reflections, and scattering from structural discontinuities such as stiffeners, which give rise to geometric mismatch, ill-posed probabilistic imaging, and localization artifacts. To address this issue, ETLM-based imaging is reformulated as a weighted geometric consistency inference problem. First, a damage index (DI) screening strategy controlled by the median absolute deviation (MAD) criterion is employed to automatically exclude damage-insensitive paths and improve the data reliability. The remaining paths are subsequently incorporated into a weighted nonlinear objective function, with weights allocated based on damage sensitivity to avoid information dilution from equal contribution. To mitigate the non-convexity induced by time of flight (ToF) uncertainty, a global meta-heuristic optimization strategy directly estimates the damage coordinates without the need for an initial guess. In addition, a Bayesian maximum a posteriori (MAP) formulation is adopted to quantify localization uncertainty and derive confidence intervals. The proposed method is validated through finite element simulations, experiments on flat CFRP laminates, and a T-stiffened CFRP laminate. These results demonstrate that it reduces localization artifacts and improves localization accuracy relative to ETLM under sparse sensor array and structurally non-uniform conditions.