A new roadmap which lays out how AI and big data techniques could drive advances in superconductivity research and development is aiming to help to spark a tech revolution.
 
An international team of leading engineers, physicists and computing scientists are behind the roadmap, which is the first of its kind for the field of superconductor research.  
 
Dr Mohammad Yazdani-Asrami of the University of Glasgow led the production of the roadmap, which is published as an invited paper in the Institute of Physics’ journal, Superconductor Science and Technology.
 
The paper showcases 18 short articles, produced by 40 researchers from 25 institutions around the world. Together, they offer a comprehensive guide to how the power of machine learning could help overcome challenges which have held back the creation of new technologies built with superconducting components.
 
Superconductors are a unique group of lightweight materials which can generate strong magnetic fields and transfer or store large amounts of energy. They are also capable of conducting electricity with zero resistance, a property which sets them apart from all other conductive materials, which lose energy as heat when current flows through them.
 
Superconductors are currently used in magnetic resonance imaging, or MRI, which has enabled major advances in medical and cancer diagnostics by creating detailed scans of the body. They have also underpinned promising advancements in particle accelerators, high-performance computing, energy storage and more.
 
In the future, new superconductor technologies could also create breakthroughs in wind power generation, fusion energy, electric and hydrogen-powered transport, and aerospace applications helping the world achieve net-zero.
 
However, a series of tough challenges have so far prevented the widespread adoption and commercialisation of superconducting technology across the full spectrum of industries. Aside from MRIs, there are currently very few superconducting devices in commercial use, with many still confined to research facilities.
 
Part of that is because industrial superconductor production is difficult, energy-intensive and expensive - an issue which is compounded by the need to cool the materials to temperatures far below zero for them to operate at peak efficiency.
 
Each section in the roadmap posits how new developments in AI and big data could help overcome problems currently holding back the development of specific areas of superconductor research.
 
The authors outline the challenges facing superconductor research in material design, manufacturing, testing, operation and condition monitoring and demonstrate how AI could help develop new approaches to solving them.
 
Those challenges include:

  • the optimal design optimization of superconducting propulsion systems in hydrogen-electric aircraft
  • fault detection in superconducting devices
  • hot spot detection in superconducting devices for fusion applications
  • real-time and surrogate modelling of superconducting systems
  • quench detection of superconducting magnets
  • new superconductor discovery
  • superconductor manufacturingA photo of Dr Mohammad Yazdani-Asrami of the James Watt School of Engineering

Dr Yazdani-Asrami, of the University of Glasgow’s James Watt School of Engineering, led the team of authors and co-ordinated the drafting and editing of the paper. He is a named author on four of its sections.
 
He said: “Superconductors have enabled some truly remarkable technologies over the last few decades and hold the promise of underpinning many more in the decades to come.
 
“Artificial intelligence and machine learning have already proven their value in many areas of science and engineering. They are invaluable for sifting through huge amounts of data, finding hidden patterns and making decisions that can help bolster human ingenuity.
 
“The roadmap is a true international collaboration, with input from experts in Europe, North and South America, Asia, and Oceania. It’s the first time that experts from a diverse range of disciplines have worked together to forecast how AI can advance cryogenic and superconducting technology towards commercialisation, which is where real change will happen.
 
“Our goal in putting the paper together was to inspire researchers from a wide range of fields to embrace the potential that AI and big data techniques have to create new opportunities for superconducting materials and technologies."
 
Dr Wenjuan Song, also from the University of Glasgow’s James Watt School of Engineering, co-authored the roadmap. She added: “The roadmap also aims to help policymakers and industry to recognise that they also have a part to play in making these breakthroughs possible.
 
“I’m excited to see how our roadmap is received by the superconductor research and development and beyond. I’m also looking forward to doing my part in driving forward superconductor advances in electrically-powered aircraft, my own field of research.”

The team’s paper, titled ‘Roadmap on Artificial intelligence and big data techniques for superconductivity’, is published in Superconductor Science and Technology and is available free of charge.


Notes to Editors: 

The sections of the roadmap are: 

1) Intelligent condition monitoring and design optimisation of superconducting propulsion machines using AI-based techniques for future cryo-electric aircraft (Mohammad Yazdani-Asrami and Wenjuan Song, University of Glasgow, UK) 

2) AI-assisted real-time modelling of HTS devices and systems (Antonio Morandi, University of Bologna, Italy, Giovanni De Carne, Karlsruhe Institute of Technology, Germany) 

3) HTS Bulk Modelling Supported by AI-Based Paradigms Roadmap (João Murta-Pina and Anabela Pronto, Centre of Technology and Systems Systems – UNINOVA (CTS-UNINOVA), Portugal; Roberto Oliveira, Karlsruhe Institute of Technology, Germany) 

4) Surrogate modelling of superconducting materials and applications (Wenjuan Song, University of Glasgow, UK, Francesco Grilli, Karlsruhe Institute of Technology, Germany, Enric Pardo, Slovak Academy of Sciences, Slovakia) 

5) AI for magnet technology and MRI industry (Michael Parizh, General Electric - Research, USA; Boyang She, and Tim Coombs, University of Cambridge, UK) 

6) Integrated magnet design environment via surrogate modelling-based optimisation (Tiina Salmi, Di Wu and Eric Coatanea, Tampere University, Tampere, Finland) 

7) Real-time hot spot detection of cryogenic fusion magnets using distributed optical fibre sensors (Dominic A. Moseley, Rodney A. Badcock, and Mengjie Zhang, Victoria University of Wellington) 

8) Detection of quench precursors in superconducting magnets using AI techniques (Vittorio Marinozzi and Nhan Tran, Fermi National Accelerator Laboratory (Fermilab), USA 

9) Development of a low-latency ML algorithm for electronic hardware for protection of superconducting magnets applied in high energy physics (Maciej Wielgosz and Andrzej Skoczen, AGH University of Science and Technology, Poland) 

10) AI and smart algorithms for the protection of SCs in modern power grids (Dimitrios Tzelepis, University of Strathclyde, UK; Mohammad Yazdani-Asrami, University of Glasgow, UK; Sakis Meliopoulos, Georgia Tech, USA) 

11) AI techniques for design development, modelling, and online monitoring of saturated core SFCLs (João Murta-Pina, and Nuno Vilhena, Centre of Technology and Systems – UNINOVA, Portugal; Guilherme Sotelo, Fluminense Federal University, Brazil) 

12) Intelligent superconducting transformers for power network and traction-transportation applications (Mohammad Yazdani-Asrami, and Wenjuan Song, University of Glasgow, UK; Zhenan Jiang, Victoria University of Wellington, New Zealand) 

13) AI and BD for improvement of the manufacturing process of superconductors (Veit Große, THEVA Dünnschichttechnik GmbH, Germany) 

14) AI applied to the development of superconducting wires with enhanced electro-mechanical properties (Tommaso Bagni, Diego Mauro, and Carmine Senatore, University of Geneva, Switzerland) 

15) Implementation of AI and BD techniques in future large scale 2G HTS tape production: SuperOx perspective (Alexey Mankevich, and Vadim Amelichev, S-Innovations, Russia; Sergey Samoilenkov, Moscow, SuperOx, Russia) 

16) Exploration of an extended Hubbard model for high-Tc superconductivity in cuprates via ML approaches (Tiem Leong Yoon, Universiti Sains Malaysia, Malaysia) 

17) Data-science enabled discovery of superconductors (Yao Wang, Clemson University, USA; Renato P. Camata, and Cheng-Chien Chen, University of Alabama at Birmingham, USA) 

18) AI for superconductivity: challenges, and future trends (Ana Maria Madureira, Polytechnic of Porto, Portugal; Mohammad Yazdani-Asrami, University of Glasgow, UK; and Ajith Abraham, Machine Intelligence Research Labs (MIR Labs), USA)

 

Researchers from the following institutions contributed to the paper:

University of Glasgow, United Kingdom

Karlsruhe Institute of Technology, Germany

University of Bologna, Italy

Centre of Technology and Systems – UNINOVA, Portugal

Slovak Academy of Sciences, Slovakia

General Electric - Research, USA

University of Cambridge, United Kingdom

Tampere University, Finland

Victoria University of Wellington, New Zealand

Fermi National Accelerator Laboratory (Fermilab), USA

AGH University of Science and Technology, Poland

University of Strathclyde, United Kingdom

The Georgia Institute of Technology, USA 

Fluminense Federal University, Brazil

THEVA Dünnschichttechnik GmbH, Germany

University of Geneva, Switzerland

S-Innovations, Russia

SuperOx, Russia

Universiti Sains Malaysia, Malaysia

Clemson University, USA

University of Alabama at Birmingham, USA

Polytechnic of Porto, Portugal

Faculty of Computing and Data Science, FLAME University, Pune, Maharashtra, India

Machine Intelligence Research Labs (MIR Labs), USA

First published: 28 February 2023