Wed, 14 July, 2021
A new paper by TWI Team Manager Channa Nageswaran, investigating the use of machine learning for the detection and sizing of fatigue cracks, has been published.
Fatigue cracks pose a significant threat to the integrity of a wide range of industrial components, structures and assets. Being able to detect these cracks with ultrasonic inspection is an economically viable answer to this problem, but the morphology of fatigue cracks can limit the effectiveness of some techniques.
As well as detecting the cracks, there is a need to measure the size of any cracking, often within the material’s volume. While ultrasonic techniques are well-suited to this, achieving the required sensitivity to the tip of the cracks is difficult. Not being able to locate the crack tip effectively creates problems with measuring the crack size.
The development of machine learning techniques are assisting with this challenge and this paper details a new approach for conditioning ultrasonic data to machine learning settings so that it can be used to effectively and confidently detect and size fatigue cracks.
The new approach, using images termed parameter-spaces, will also find use in conventional inspections as they are able to provide operators with information as to the existence (or not) of these dangerous cracks.
The developments highlighted in this paper will lead to further innovations in the way future ultrasonic techniques are implemented for industrial inspection.
You can read the paper, ‘The Snooker Algorithm for Ultrasonic Imaging of Fatigue Cracks in order to use Parameter-Spaces to Aid Machine Learning,’ by following this link.