Machine Learning promises improved safety analysis

By May 4, 2020Nuclear

Sellafield’s commitment to constantly improving working practices and conditions across the site may find that a data analytics technique called Machine Learning (ML) holds a key to success.

ML teaches computers to learn from experience and is widely used in many technologies which impact daily life including mobile phone apps and internet search engines.  The technique is applied in a broad range of sectors including financial services, manufacturing and healthcare to improve decision making.

And now Sellafield wants to explore how ML could be used to analyse many thousands of safety reports covering a considerable number of years. The organisation want to be able to scan the data for trends, helping transform insight and early prediction, and thereby prevention, of safety issues. Currently the data is stored in Sellafield’s document systems and interrogated manually – a laborious and time consuming process.

Through the Game Changers innovation programme, Sellafield issued a challenge to any organisation or individual from any sector who could offer a potentially viable solution to the challenge.

Five successful groups were chosen from the 30 proposed solutions and will receive up to £10k each to develop the feasibility of their ML ideas.

Successful organisations in the initial stage of the ML challenge are Assystem, Spotlight Data, National Nuclear Laboratory, the Health and Safety Executive and Ryelore Ai. Further proof of concept funding may be awarded by Game Changers depending on feasibility success.

Andrew Cooney, Decommissioning Technology Manager at Sellafield said: “At a time when everyone is working in difficult circumstances during the Coronavirus pandemic, it’s great that Game Changers can continue their normal business by backing these organisations to develop these important projects for Sellafield.”