Development and assessment of modified total focusing method image reconstruction techniques based on guided waves for aluminium and composite health monitoring

UoM administered thesis: Phd

  • Authors:
  • Aurelia Muller


This research focuses on the development of image reconstruction techniques for the structural health monitoring (SHM) of aluminium and composite structures with guided waves. Structural health monitoring systems are promising for reducing maintenance cost in the aerospace industry. However, most systems still lack the maturity and high reliability required to meet the demanding standards. In the past decade, several studies have adapted and applied image reconstruction techniques, in particular the total focusing method (TFM) imaging algorithm, to guided wave signals. However, a review of the literature has identified inconsistencies in the defect location accuracy of the tested TFM methods and few attempts to characterise the detected defect in the reconstructed images. This experimental research contributes to the understanding of the effect of several modifications to the TFM image reconstruction process on the system’s ability to detect and accurately locate defects. The modifications include using two different baselines and normalising or not normalising the signals, as well as a novel modification that involves a selective process of only the positive or the negative amplitudes of the residual signals. These modifications lead to 12 TFM image reconstruction methods, which were tested on signals collected in aluminium and carbon/epoxy composite plates. The plates were fitted with a circular array of piezoelectric disk transducers and subjected to successive damage states involving multiple hole- and crack-type defects of various sizes. A set of metrics, including distance from expected position and contrast to noise ratio, were used to compare the performance of the TFM methods. In addition, numerical quantification of the defect shapes in the images, including shape segmentation, circularity factors, eccentricity factors, and area ratio, were proposed. This study found that images reconstructed using exclusively positive amplitude in the residual signals were the most accurate and consistent in locating defects in both the composite and aluminium plate. In contrast, composite plate images reconstructed using exclusively negative amplitude had their location accuracy significantly decreased when multiple defects were present. In addition, analysis revealed that the smallest induced defects in the plates had their areas systematically overestimated in the reconstructed image. Furthermore, a high segmentation threshold could lead to underestimating the size of larger induced defects. Finally, the normalisation of signals increased the probability of detecting and locating defects when reducing the number of transducers.


Original languageEnglish
Awarding Institution
Award date31 Dec 2019