Machine learning for real-time inversion of locally anisotropic weld properties using in-process ultrasonic array measurements

Research output: Contribution to conferenceSpeechpeer-review

Abstract

Many welds exhibit significant local anisotropy due to their elongated grain structure. Ultrasonic inspection of these anisotropic materials is challenging as the variations in sound speed alter the direction of acoustic propagation within the component. This is extremely detrimental to the quality of ultrasonic array imaging, as without accurate knowledge of the local anisotropic properties, precise focusing cannot be achieved. Currently, the common approach to remedy this issue is to use lower frequencies (This paper aims to improve ultrasonic array based inspection of anisotropic welds by mapping local variations in anisotropy. This is achieved by training a neural network to invert for the grain orientations in the weld using time-of-flight (TOF) measurements. 64x64 TOF matrices are measured using two arrays, in tandem, positioned either side of the weld. To maximize the information available to the neural network this measurement is taken ‘in-process’, after each layer of weld material deposition. The neural network is trained on data simulated using the anisotropic locally interpolated fast marching method (ALI-FMM) (Ludlam et al., 2023). Once trained, the neural network is tested with both finite element and experimental data from a 316L stainless steel weld. Performance is measured in two ways. Firstly, by comparing the predicted and true grain orientations where this ground truth is available, and secondly, by using the predicted orientations to calculate anisotropic travel time maps and observing the improvement in image quality compared to imaging with an assumption of isotropy. The feasibility of this approach has been previously demonstrated in (Singh et al., 2022) for a simplified simulated inspection.

Ludlam, J., Dolean, V., & Curtis, A. (2023). Travel times and ray paths for acoustic and elastic waves in generally anisotropic media. ArXiv Preprint ArXiv:2302.10988.
Singh, J., Tant, K., Mulholland, A., & MacLeod, C. (2022). Deep learning based inversion of locally anisotropic weld properties from ultrasonic array data. Applied Sciences, 12(2), 532.

Original languageEnglish
Publication statusPublished - 12 Sept 2023
Event60th Annual British Conference on NDT - Northampton Town Centre Hotel, Northampton, UK , Northampton, United Kingdom
Duration: 12 Sept 202314 Sept 2023
https://www.bindt.org/events-and-awards/ndt-2023/

Conference

Conference60th Annual British Conference on NDT
Abbreviated titleBINDT 2023
Country/TerritoryUnited Kingdom
CityNorthampton
Period12/09/2314/09/23
Internet address

Keywords

  • Machine Learning
  • anisotropic weld properties
  • ultrasonic inspection

Fingerprint

Dive into the research topics of 'Machine learning for real-time inversion of locally anisotropic weld properties using in-process ultrasonic array measurements'. Together they form a unique fingerprint.

Cite this