No abstract

No abstract

Many researchers have proposed vibration-based damage-detection approaches for continuous structural health monitoring. Translation to real applications is not always straightforward because the proposed methods have mostly been developed and validated in controlled environments, and they have not proven to be effective in detecting real damage when considering real scenarios in which environmental and operational variations are not controlled. This work was aimed to develop a fully-automated strategy to detect damage in operating tie-rods that only requires one sensor and that can be carried out without knowledge of physical variables, e.g., the axial load. This strategy was created by defining a damage feature based on tie-rod eigenfrequencies and developing a data-cleansing strategy that could significantly improve performance of outlier detection based on the Mahalanobis squared distance in real applications. Additionally, the majority of damage-detection algorithms presented in the literature related to structural health monitoring were validated in controlled environments considering simulated damage conditions. On the contrary, the approach proposed in this paper was shown to allow for the early detection of real damage associated with a corrosion attack under the effects of an intentionally uncontrolled environment.

Axially loaded beam-like structures represent a challenging case study for unsupervised learning vibration-based damage detection. Under real environmental and operational conditions, changes in axial load cause changes in the characteristics of the dynamic response that are significantly greater than those due to damage at an early stage. In previous works, the authors proposed the adoption of a multivariate damage feature composed of eigenfrequencies of multiple vibration modes. Successful results were obtained by framing the problem of damage detection as that of unsupervised outlier detection, adopting the well-known Mahalanobis squared distance (MSD) to define an effective damage index. Starting from these promising results, a novel approach based on unsupervised learning data clustering is proposed in this work, which increases the sensitivity to damage and significantly reduces the uncertainty associated with the results, allowing for earlier damage detection. The novel approach, which is based on Gaussian mixture model, is compared with the benchmark one based on the MSD, under the effects of an uncontrolled environment and, most importantly, in the presence of real damage due to corrosion.

No abstract