Vibration-based damage detection in beam-type structures by using a Bayesian data fusion based technique

Abstract

The study in the field of damage detection in civil, marine, mechanical and aerospace engineering structures has always acquired great interest in order to avoid risks related to ageing of the structure and linked possibility for damage accumulation. Therefore health monitoring is becoming ever more important for these structures. Numerous non-destructive damage identification techniques are available which are highly effective on local basis. However, damage being typically unknown in both presence and location, make them ineffective especially for large and complex structures. Vibration-based damage detection (VBDD) is an active research area, which as compared to local non-destructive evaluation methods, does not require prior knowledge about the damage location. Most of these methods are model-based which requires information from either the undamaged structure itself or to compare with a numerical model which makes the methods inflexible. However a number of techniques are available which consider the damaged structure only, to identify damage but the unavoidable noise in the measurements creates a detrimental effect on these techniques. The research in this thesis presents a noise reduction technique with the help of which a study is done to evaluate two conventional VBDD methods i.e., Gapped-Smoothing method (GSM) and Modal Strain Energy method (MSEM) in beam-type structures, without using the baseline information about the structure. The developed method uses the damage indices provided by GSM and MSEM for synthesizing a set of likelihood function that is processed under a Bayesian approach in order to reduce the effect of the noise and other uncertainty sources. The quality of the damage detection for the beams under consideration is examined by investigating optimal sampling size analytically and then through numerical simulation and experiments. The study comprises of testing a steel beam with multiple transverse edge cracks and a glass fibre reinforced plastic (GFRP) beam with a through-width delamination. It is demonstrated that by using a suitable sample size, the developed method can be successfully employed to detect the location of the damage using both GSM and MSEM without any baseline information. Furthermore, it is observed that as compared to MSEM, the developed method with GSM works better on edge cracks. Conversely, the developed method with MSEM provides better results for the localisation of delamination than with GSM.