Thesis

Automated ultrasound data processing for defect detection and characterisation through machine learning

Downloadable Content

Download PDF
Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2025
Thesis identifier
  • T17467
Person Identifier (Local)
  • 202170218
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • The growing adoption of Carbon Fibre Reinforced Plastic (CFRP) composites insafety-critical structures, such as aircraft fuselages and wind turbine blades, requiresthorough inspection process to ensure material integrity and prevent catastrophicfailures. These inspections are conducted using Non-Destructive Evaluation (NDE), acollective term for methods that assess the quality of materials without causingdamage. Among these methods, Ultrasonic Testing (UT) stands out as a preferredinspection technique in the aerospace industry. The field of NDE has experiencedsignificant advancements with the introduction of advanced sensor technologies androbotic manipulators, which have automated and accelerated data acquisitionprocesses. However, data analysis and interpretation remain predominantly manual,making the process time-consuming and prone to human error, thus creating abottleneck in manufacturing. Recent advancements in Artificial Intelligence (AI)present new opportunities to automate these tasks.This thesis explores the application of AI techniques to analyse UT datasets obtainedfrom reference CFRP samples representative of those used in the aerospace industry.The initial research focused on evaluating the performance of supervised AI methods,specifically object detection models, in defect detection tasks using ultrasonic C-scanimages, which represent the top cross-sectional view of the inspected materials. As abaseline, both a traditional signal thresholding technique and an enhanced statisticalthresholding method, based on theoretical mathematical distributions fitted to theobserved data, were examined. The primary contribution of this work is thedemonstration of the superior performance of AI models in this context overthresholding methods. Additionally, supervised training was conducted exclusively onsimulated data, thereby addressing the data scarcity challenge.Building on this, an unsupervised AI method in the form of anomaly detection wasexplored. This approach addressed the scarcity of datasets containing defectiveindications and the challenges of relying on large-scale simulations, which requiresignificant computational resources and extensive manual effort to generate groundtruths for supervised training. A two-step workflow was developed, comprising anautomated signal gating method based on unsupervised clustering and an autoencoder model serving as an anomaly detector applied to ultrasonic B-scans (which representcross-sectional images of the material). The key advantages of this method include astreamlined development process focused on the use of pristine data (in this thesis, theterm pristine refers to samples that contain no intentional or unintentionalmanufacturing defects that are detectable using the ultrasonic inspection setupemployed, within the limits of its resolution), which is more readily available, and theelimination of ground truth generation requirements. This workflow was successfullyapplied to samples with both uniform thickness and complex geometries. Additionally,this research investigated the impact of human factors on AI results and highlightedthe challenges posed by inconsistent data quality during scans.The final stage of this research focused on strategies for integrating the developed AImodels into NDE data analysis, driven by the recent evolution towards NDE 4.0 (aconcept that combines digitalisation, automation, and connectivity to modernise NDE)focused on enabling fully automated systems. Despite significant research in this field,implementation strategies are often underexplored, and clearly defined automationlevels achievable with AI remain lacking. To address these gaps, four levels of dataanalysis automation using AI were defined and evaluated. Additionally, thesynchronous use of multiple AI models, each designed to process distinct views of theultrasonic data, enabled cross-validation between models. This approach enhancedtrust in the automated system while offering mechanisms to mitigate potentialperformance degradation of NDE operators using such systems. Furthermore, thisstrategy aligns with NDE 4.0 objectives, transitioning operators into supervisory roleswhile delegating repetitive tasks to the AI systems. The developed methods wereevaluated in a case study involving complex geometry samples, demonstrating theireffectiveness for potential industrial applications.Overall, this thesis proposes methods to assist in automating NDE data analysisprocesses for UT of CFRP composites, with the primary goal of enhancing accuracyand reducing analysis time to address the current bottleneck in aerospacemanufacturing.
Advisor / supervisor
  • Mohseni, Ehsan
Resource Type
DOI
Funder

Relations

Items

Thumbnail Title Date Uploaded Visibility Actions
file details PDF of thesis T17467 2025-10-15 Public Download
file details File 2025-10-15 Private