The Turing has been awarded the Hojjat Adeli Award for Innovation in Computing for the paper Hierarchical Bayesian modeling for knowledge transfer across engineering fleets via multitask learning.
The annual award, established by the publishers Wiley-Blackwell, is given to the most innovative computing paper published in the previous volume or year or by the most groundbreaking single author.
In this work, the researchers show that models which analyse how natural ecosystems behave and react to their environment can also offer insights in engineering. Machines, just like plants and animals, may have shared characteristics but no two will be identical and their behaviour depends on their environment.
The researchers propose analysing machines as an interconnected ecosystem, rather than as individual components. They believe this could offer new insights and help address the problem of data sparsity when predictive models are being built for engineering infrastructure.
However, this type of approach is much less developed for machines and structures. So this paper aims to better understand how machines behave to help make them more efficient.
The research looks at two case studies as potential applications of this approach. The first considers the survival analysis of components in an operational fleet of trucks and the second predicts the power in a group of wind turbines.
This paper includes some of the research undertaken as part of the Turing’s AI for Science and Government (ASG) programme. In particular, research which took place within the theme of Ecosystems of Digital Twins which is referenced in the digital twins white paper.
The paper was a collaborative effort involving teams from the University of Cambridge, The Alan Turing Institute, the University of Glasgow, and Scania, an industry partner, Imperial College London and Stockholm University.
Professor Mark Girolami, Chief Scientist at The Alan Turing Institute, said: “We are delighted that our work has been recognised by receiving this prestigious award. By examining machines as part of an interconnected ecosystem, as one would study natural populations, we can gain new insights into their emergent behaviour and performance. In addition, our work demonstrates how approaching a problem in a collaborative and innovative way, can help give us new ideas for improving systems.
