UC San Diego

Nondestructive Evaluation (NDE) and Structural Health Monitoring (SHM) using ultrasonic waves have become compelling techniques for identifying various defects in structural components. Ultrasonic guided-wave testing has the potential to identify the elastic properties of the fiber-reinforced composite laminates. A nondestructive tool for characterization of the composite properties is particularly advantageous when components involve manufacturing variances and service quality degradation. The first part of this dissertation examines the potential for composite property characterization by a single wave propagation direction to enable the accurate identification of several elastic properties away from the wave propagation direction because of the anisotropy of the composite providing the “coupling” effect. A property inversion scheme was proposed based on matching phase velocity dispersion curves of relevant guided modes by means of a Simulated Annealing optimization algorithm and a Semi-Analytical Finite Element method to solve the forward problem. The second part of the dissertation is focused on defect detection and localization in a stiffened skin-to-stringer composite panel. A data-driven Deep Learning approach based on Convolutional Neural Network is exploited to detect the damages even generalized to non-training scenarios. Moreover, a matched-field-data-driven method and the structural transfer function method (the deconvolution of the “dual-output” scheme) are also discussed to explore a wide field for the damage detection of the skin-to-stringer assembly. The third part of this dissertation targets defects imaging on homogeneous solids using ultrasonic bulk waves to reconstruct 2D and 3D images of defects. Sensor arrays and a frequency-domain-based beamforming algorithm are employed to improve the defect characterization process. The experimental applications are performed on both a simulated defect on an aluminum block in the lab and natural transverse defects on the railroad tracks at the Transportation Technology Center Inc (TTCI) in Pueblo, CO. Moreover, a preliminary study of the fatigue history prediction of the railway track is also discussed for future investigation.