Picture water droplets

Developing mathematical theories of the physical world: Open Access research on fluid dynamics from Strathclyde

Strathprints makes available Open Access scholarly outputs by Strathclyde's Department of Mathematics & Statistics, where continuum mechanics and industrial mathematics is a specialism. Such research seeks to understand fluid dynamics, among many other related areas such as liquid crystals and droplet evaporation.

The Department of Mathematics & Statistics also demonstrates expertise in population modelling & epidemiology, stochastic analysis, applied analysis and scientific computing. Access world leading mathematical and statistical Open Access research!

Explore all Strathclyde Open Access research...

Advanced defect detection algorithm using clustering in ultrasonic NDE

Gongzhang, Rui and Gachagan, Anthony (2016) Advanced defect detection algorithm using clustering in ultrasonic NDE. AIP Conference Proceedings, 1706. ISSN 0094-243X

[img]
Preview
Text (Gongzhang-Gachagan-AIP-2016-defect-detection-algorithm-using-clustering-in-ultrasonic-NDE)
Gongzhang_Gachagan_AIP_2016_defect_detection_algorithm_using_clustering_in_ultrasonic_NDE.pdf
Final Published Version

Download (512kB) | Preview
[img]
Preview
Text (QNDE 2015 - Advanced Defect Detection Algorithm Using Clustering in Ultrasonic NDE)
QNDE_2015_Advanced_Defect_Detection_Algorithm_Using_Clustering_in_Ultrasonic_NDE.pdf
Accepted Author Manuscript
License: Unspecified

Download (522kB) | Preview

Abstract

A range of materials used in industry exhibit scattering properties which limits ultrasonic NDE. Many algorithms have been proposed to enhance defect detection ability, such as the well-known Split Spectrum Processing (SSP) technique. Scattering noise usually cannot be fully removed and the remaining noise can be easily confused with real feature signals, hence becoming artefacts during the image interpretation stage. This paper presents an advanced algorithm to further reduce the influence of artefacts remaining in A-scan data after processing using a conventional defect detection algorithm. The raw A-scan data can be acquired from either traditional single transducer or phased array configurations. The proposed algorithm uses the concept of unsupervised machine learning to cluster segmental defect signals from pre-processed Ascans into different classes. The distinction and similarity between each class and the ensemble of randomly selected noise segments can be observed by applying a classification algorithm. Each class will then be labelled as 'legitimate reflector' or 'artefacts' based on this observation and the expected probability of defection (PoD) and probability of false alarm (PFA) determined. To facilitate data collection and validate the proposed algorithm, a 5MHz linear array transducer is used to collect A-scans from both austenitic steel and Inconel samples. Each pulse-echo A-scan is pre-processed using SSP and the subsequent application of the proposed clustering algorithm has provided an additional reduction to PFA while maintaining PoD for both samples compared with SSP results alone.