Guided Wave-based Structural Health Monitoring (GWSHM) systems are used to monitor and assess the integrity of structures like pipelines, power plants, and aircraft non-destructively by permanently mounting transducers and automatically evaluating the recorded data. Even though having such systems increases the level of safety in life, the reliability assessment of GWSHM systems is a vital burden that prevents them from being widely deployed in more industrial applications as there is no clear guideline to assess their reliability. Early considerations suggested that the reliability assessment of SHM systems should be based on a pre-existing guideline like the Non-Destructive Evaluation (NDE) metric such as parametric Probability of Detection (POD) analysis, known as classical POD analysis. However, translating the parametric POD analysis precisely, as described in MIL-HDBK 1823A, is not applicable due to the statistical dependence of repeated measurements and the inability to use the considered structure after intentionally damaging it. The Model-Assisted Probability of Detection (MAPOD) technique emerges as a promising approach that enables the generation of a statistically independent dataset using numerical simulation models only. The same procedures as the classical parametric POD analysis, described in MIL-HDBK 1823A, are used to assess the reliability of the GWSHM system. Although simulation tools are highly effective in accounting for and simulating structural uncertainties, they fall short in modelling system, operational and environmental uncertainties. Moreover, the resulting POD analysis remains simulation-based, lacking seamless integration with reality. Above all, the classical parametric POD analysis is confined by specific guidelines mandating a linear relationship between damage size a and its corresponding damage indicator for applicability. This thesis presents an innovative method called the 'Augmented POD method', which addresses these issues by combining features extracted from experimental datasets taken from a real GWSHM system in its pristine state under environmental and system uncertainties with different 2D and 3D damage scenarios modelled using the Pogo software commercial tool. The Augmented POD (A-POD) method applies to any relationship between the damage indicator and damage size a. It is suitable for both linear and non-linear relationships. The A-POD method is demonstrated on a steel pipe with a GWSHM system employing an array of transducers bonded around the circumference of the pipe to excite the fundamental pure torsional mode. Different 2D and 3D damage growths are simulated, considering the structure uncertainties. Then, an artificial damage is introduced to the pipe to investigate the alignment between the simulation and the experimental damages. The resultant output of the novel A-POD method is higher than that of a purely simulation-based POD analysis due to the incorporation of realistic uncertainties.
Ahmed Bayoumi (Thu,) studied this question.