Optimal methodologies for ultrasonic guided-wave based structural health monitoring

Cantero Chinchilla, Sergio (2020) Optimal methodologies for ultrasonic guided-wave based structural health monitoring. PhD thesis, University of Nottingham.

[img]
Preview
PDF (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Available under Licence Creative Commons Attribution.
Download (17MB) | Preview

Abstract

The assessment of structural integrity is a key issue for many industries due to its important implications in safety, maintenance cost reduction, and improved asset availability. In this context, structural health monitoring (SHM) systems using ultrasonic guided-waves are being explored for an efficient diagnosis of damage and prognosis of the remaining useful life of the monitored structure. Nonetheless, addressing this monitoring scenario is a challenge given the inherent complexities associated to each of the diagnosis steps, which encompass the optimal SHM design, the detection of damage, its localisation, and its identification. Among these complexities, uncertainties stemming from several sources such as equipment noise, manufacturing defects, and the lack of conclusive knowledge about wave propagation introduce a high variability in the response of the SHM system. The main objective of this thesis is to provide probabilistic Bayesian and fuzzy logic methodologies to manage global uncertainties for each step in the SHM process.

The accuracy and reliability of an ultrasonic guided-wave based SHM system are dependent on the chosen number and location of sensors and actuators. A general framework for optimal sensor configuration based on value of information is proposed in this thesis, which trades-off information gain and cost. This approach optimally chooses the sensor position so that they render the largest information gain when inferring the damage location. The methodology is tested using different case studies in the context of ultrasonic guided waves and piezoelectric sensors. However, although this framework is mathematically rigorous, it is computationally expensive should the actuators be considered in the optimisation problem. To overcome this issue, a cost-benefit analysis is also proposed using both the Shannon's information entropy and a cost function associated to the number of sensors and actuators. The objective function is based on binary decision variables, which are relaxed into continuous variables, hence convexifying the objective function. This optimisation methodology is illustrated in several case studies considering plate-like structures with irregular geometries and different materials, providing a high computational efficiency.

The first diagnosis stage requires a robust and computationally efficient damage detection approach in real-life engineering scenarios. To this end, a novel damage index for ultrasonic guided-wave measurements based on fuzzy-logic principles is proposed in this thesis. This approach assesses the time of flight mismatch between signals acquired in undamaged and non-pristine states using fuzzy sets for its evaluation. The robustness partially builds on the use of a large amount of signals stemming from two experimental procedures: the round robin configuration and the transmission beamforming technique. This new damage index is validated in several scenarios with sudden and progressive damage.

Once a damage area has been detected, the next diagnosis stage requires a reliable damage localisation. To address this SHM step, a robust methodology is proposed based on two hierarchical levels: (1) a Bayesian time-frequency model class selection to obtain the time of flight of damage scattered waves; and (2) a Bayesian inverse problem of damage localisation that considers as input data the outcome of the first level. The effectiveness and robustness of the proposed methodology is illustrated using two cases studies with one and two areas of damage.

Lastly, to provide a complete diagnosis of damage using ultrasonic guided-waves, the identification of damage needs to be addressed. A multi-level hybrid wave and finite element model-based Bayesian approach is proposed to identify the type of damage in composite beams based on posterior probabilities, hence accounting for different sources of uncertainty. In addition to the type of damage, this approach allows the inference of damage-related parameters and the damage location. A carbon fibre beam with two damage modes, i.e. a crack and a delamination, is used to illustrate the methodology.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Jones, I.A.
Chronopoulos, Dimitrios
Chiachío Ruano, Juan
Keywords: Structural health monitoring, Optimal sensor placement, Value of information, Bayesian inverse problem, Ultrasonic guided waves
Subjects: T Technology > TA Engineering (General). Civil engineering (General) > TA 630 Structural engineering (General)
Faculties/Schools: UK Campuses > Faculty of Engineering
UK Campuses > Faculty of Engineering > Department of Mechanical, Materials and Manufacturing Engineering
Item ID: 61362
Depositing User: Cantero Chinchilla, Sergio
Date Deposited: 31 Dec 2020 04:40
Last Modified: 31 Dec 2020 04:40
URI: https://eprints.nottingham.ac.uk/id/eprint/61362

Actions (Archive Staff Only)

Edit View Edit View