PhD Studentship – Artificial Intelligence Methods – Automated Defect Recognition of Ultrasonic Data in NDT
Funding Amount: UK/EU tuition fees plus competitive stipend
Regions: EU (Non UK), UK
Duration: 3 years
Application deadline: 4th October 2019
The School of Engineering at the University of Wales Trinity St David and TWI Ltd are seeking a top class candidate to undertake research leading to the award of a 3 year PhD awarded by the University of Wales Trinity St David. As a Doctoral student you will have unrivalled access to some of the best teaching and research facilities in the world. During the 4 years you will spend approximately 9 months on advanced technical and professional development courses – usually spending the first year at University of Wales Trinity St David and the remaining time at TWI Technology Centre, Wales.
Non-Destructive Testing (NDT) is used across all industry sectors in order to assist in structural integrity assessments of safety critical assets. In particular, ultrasonic testing is the most far reaching and commonly used inspection method for sub-surface inspection, due to its portability, high probability of flaw detection, flaw sizing accuracy, affordability of equipment and avoids health and safety issues arising from ionising radiation associated with radiographic techniques. However, ultrasonic data can be difficult to interpret and signals which may appear to be flaws can in fact be an artefact arising from geometry or simply material noise.
The use of Machine Learning / Artificial Intelligence (ML/AI) is now commonplace among a wide variety of industry sectors, including computer gaming, marketing and security. While some work has been carried out for NDT, there are a lack of commercially available systems that exploit the technology. Primarily this is due to the complex nature of NDT components, inspection parameters and obtainable data. Consequently, there is a real need to determine the applicability of a range of AI techniques for use in ultrasonic inspection and for NDT in a wider context.
a) what is the specific research question being addressed:
This project will investigate which AI approach is most applicable for a range of ultrasonic inspection methods. The proposed research will seek to establish a solid understanding of available AI methods establish their suitability for NDT based on academic rigour and real-world trials. Flexible AI algorithms should be developed for use with ultrasonically acquired data. The benefits and limitations of the chosen approaches should be assessed through a parametric study on industrial samples. Although AI methods can possibly be used with a wide variety of NDT techniques, the scope of this project is to consider ultrasonic inspection for the well-defined girth-weld inspection scenario only. This is to ensure parameter inputs are minimised while techniques and algorithms are developed.
b) what methodologies / approaches / techniques will be used in the project:
The use of artificial intelligence has shown tremendous breakthroughs in a number of areas including self-driving cars, speech synthesis, the digital entertainment industry and facial recognition. These high profile successes have come about due to an increase in computer processing power, but also due to the increased availability of open-source machine learning frameworks such as TensorFlow or Caffe. The availability of large datasets and open software architectures allow researchers to quickly tweak and optimise a neural network to a specific application. This project aims to use a number of open and readily available AI techniques to perform defect detection and classification on ultrasonic NDT data.
c) What is the role of the partner:
TWI Technology Centre Wales are a major international research and development institution providing high end applied NDT research, development and testing to a wide range of global industries and clients. UWTSD have formed a strategic working relationship with TWI over a number of years including a range of MSc and PhD research projects. TWI are able to provide the ideal environment for the proposed research programme.
d) what are the specific outputs form this research project:
Review of Current AI Techniques and their Applicability to NDT. This is to include an extensive literature review into the ultrasonic background theory to determine the suitability of mainstream computer vision and other AI techniques existing in non NDT related industry sectors (such as medical, audio and visual). During this time it is expected that some initial development of computer code is undertaken to acquire and process ultrasonic data, where the existing AI algorithms are applied to ultrasonic data to assess their suitability. As NDT deals with sentencing of real-world parts, this portion of the project will also look at how AI fits into the current regulatory environment for testing and evaluation.
Development of AI Algorithms for use with Ultrasonic Data. Running concurrently with the literature review, it is expected that the PhD student will begin to determine suitable AI approaches for the analysis of ultrasonic data and the subsequent identification and classification of defects. The student is expected to begin to develop or modify existing algorithms to test their usefulness for NDT. It is expected that real-world industrial samples will be available including reference samples, and part of this task will be data gathering in a sensible manner to establish a repository of real-world training and test data.
Development of Flexible Modelling and Simulation Capability. Part of determining the most suitable AI approach for use with ultrasonic data will be to determine which input parameters have the largest influence on the final decisions taken by the AI. Most industrial components are manufactured to high standards and rarely contain defects. To be able to obtain sufficient data for training, a modelling capability for simulating defect responses will be developed. It is envisioned that the simulated data will be combined with real-world data to further train and optimise any developed AI algorithms.
Parametric Study on Industry Provided Samples. Once the outcomes of the first three tasks have been completed, a critical assessment of the use of AI for evaluating ultrasonic NDT data is to be undertaken as a parametric study. The definition of input parameters and their effect on overall
performance will be evaluated. Industry provided samples will form a new dataset to allow the student to assess the benefits and limitations of the developed AI techniques.
UWSTD has an active Non Destructive Testing (NDT) research group, which is unique in Wales, consisting of a number of ongoing funded collaborative R&D projects with industrial partners. The University offers the only dedicated NDT postgraduate programme in Europe with a total of 10 PhDs and 21 MSc research dissertations being successfully completed with industrial partners since 2007. We presently have a growing number of fully funded PhD industrial studentships (7 at present) with the 3 industrial partners. Research has been undertaken on both an industrial and academic basis with particular strength residing in the advanced ultrasonic array signal and image processing and electromagnetic methods research.
TWI Technology Centre Wales is a regional office of TWI Ltd. It is a state of the art facility offering research and development in cutting edge inspection technologies vital to industrial structural integrity management. The aim is to ensure that industry is provided with safe, reliable and cost effective solutions and services. Located in Port Talbot, TWI Wales is within easy reach of Cardiff, (Wales’ capital City), Swansea City, as well as the Gower and Pembrokeshire peninsulas.
Candidates should have a relevant degree at 2.1 minimum, or an equivalent overseas degree, in engineering, physics, computing or a related subject. Overseas applicants should also submit IELTS results (minimum 6.5) if applicable. Good team-working, observational and communication skills are essential. Previous experience of electronics design, fabrication and build would be advantageous, however training and hands-on experience will be provided. The project will require occasional travel on a short-term basis within the UK and overseas. A Baseline Personnel Security Standard (BPSS) organised by TWI must be achieved under the condition of this offer.
Open only to those within the KESS East Wales Programme Area. Candidates must meet the following criteria:
· Have a home/work address in the East Wales Programme Area * at the time of their application;
· Have the right to take up paid work in the Programme Area * on completion of the scholarship;
· Be classified as ‘home’ or ‘EU’ according to the University’s guidelines.
* East Wales Programme Area includes: Flintshire, Powys, Wrexham, Vale of Glamorgan, Monmouthshire, Newport, and Cardiff
Funding: Funded by KESS East. Knowledge Economy Skills Scholarships is a pan-Wales higher level skills initiative led by Bangor University on behalf of the HE sector in Wales. It is part funded by the Welsh Government’s European Social Fund (ESF) convergence programme for West Wales and the Valleys.
This studentship consists of full UK/EU tuition fees at UK/EU rate, as well as a tax-free Doctoral Stipend starting at £14,628 p.a. for 2019/20. This will be topped up by TWI to a circa £20K pa for the first year rising to £24K as part of an initiative known as AEMRI (Advanced Engineering Materials Research Institute), which is funded by the Welsh European Funding Office (WEFO) using European Regional Development Funds (ERDF).
Each scholarship has an additional budget for travel, equipment/consumables and training to support your research. KESS PhD Scholarship holders do not pay fees.
Length: Full-time for 3 years. (Theses must be submitted 6 months after the funded three-year study period.)
Training: The achievement of a Postgraduate Skills Development Award (PSDA) is compulsory for each KESS scholar (The PSDA is based on a 60-credit award, which is an additional award to the PhD).
How to apply: A completed application package should be submitted to Dr Matt Briggs firstname.lastname@example.org by the 4th October 2019. Applications received after this point will not be considered. Hard copy may be sent to
Dr Matt Briggs
Research, Innovation and Enterprise Services
University of Wales, Trinity Saint David.
J-Shed, Suit 10 / 11.
Shortlisted candidates will be informed 10th October 2019
Interviews will be held w/c 14th October 2019
Please download the following documents from the UWTSD website here http://www.uwtsd.ac.uk/rdp/funding-and-projects/kess-ii/
- KESS II Participant Application Form
- Supporting Documents mentioned on each form
To discuss the project please contact: TWI – Nick Couling email@example.com UWTSD – Peter Charlton – firstname.lastname@example.org
To discuss the application process, please contact Dr Matt Briggs. email@example.com