This thesis examines wavelets and their use in processing signals obtained from non-destructive testing, a technique used in the aerospace industry to determine the integrity of a structure without requiring complete access. Specifically, this thesis investigates the applicability of wavelet processing to guided wave testing for damage detection and characterisation in beams.
A simulation model of guided wave propagation was developed, including the interaction between waves and material defects in the beam. The model combines the basic physics of wave propagation with a model of damage as a region of inhomogeneity within the beam. From this model, a tool was developed which generates theoretical signal response to guided waves at a measurement location on the beam.
The model produced responses that match well to experimental data for guided wave testing, though failed to accurately model dispersion and coupling of longitudinal and flexural modes. The model was used to produce responses for a number of different beam, damage and measurement configurations. These responses were then utilised to conduct discrete, stationary and continuous wavelet analysis and to examine the effect of varying wavelet parameters.
Wavelet post-processing of the generated responses was found to have little improvement in resolving damage signatures where interference with beam boundaries occurred. Significant improvement in feature resolution was provided in cases where noise obscured part or all of a damage signature.
Discrete and continuous wavelet transform processing was found to decrease the lower bound of detectable flaw size when restricted by noise, offering improved probability of detection and characterisation of damage in beams.