A comparison and extensions of algorithms for quantitative imaging of laminar damage in plates. II. Non-monopole scattering and noise tolerance

A generalisation of diffraction tomography that is applicable for mixed-mode Lamb wave imaging is derived, subject to clearly stated simplifying assumptions. The resulting algorithm is further extended to near-field imaging. This algorithm requires a knowledge of the scattering pattern for an appropriate point scatterer, which needs to be determined separately for each type of damage, and for every combination of incident and scattered wave modes. The procedure for determining this point-scatterer pattern is presented for the case of delamination damage modelled as a flexural inhomogeneity. The performance and limit of validity of the resulting algorithm for the A 0 mode are shown to be comparable with those for the acoustic model, but this excellent performance is accompanied by an increased sensitivity to noise relative to simpler algorithms based on beamforming or time reversed imaging. Several strategies for enhancing the noise tolerance are proposed and investigated. Experimental results are presented to demonstrate the practical implementation and imaging performance. These results also illustrate the use of the distorted wave Born approximation for imaging in the presence of known structural complexities, as well as the use of a factorisation approach to minimise the required number of sensors