Skip to main content

Perfecting industrial processes: scientists fuse human knowledge and technological data – University of Reading

Show access keys

Perfecting industrial processes: scientists fuse human knowledge and technological data

Release Date 25 April 2006

Mass manufacturing of products, such as aluminium plate, requires millions of pounds worth of investment every year, and aluminium plate producers naturally seek to maximise the productivity and profitability of rolling mills. Now, scientists at the University of Reading, in conjunction with the University of Leicester and Alcoa, have developed a new state-of-the-art 'Fused Expert System' that has shown through plate rolling trials how mills could work at optimum performance levels. Dr Will Browne, of Reading's School of Systems Engineering, together with Professor Ian Postlethwaite and Dr Liqun Yao (Leicester), created the system that fuses the human and computer knowledge of a rolling mill and uses that combined understanding to produce high quality plates of aluminium – potentially saving manufacturers millions of pounds. The system is described in the April 2006 issue of Engineering Applications of Artificial Intelligence. "The UK is one of the world leaders in metals production, but, for supervisory and plant-wide control systems, it had lost ground by failing to embrace new technologies," said Professor Postlethwaite. "This project redressed this situation through the development of a generic knowledge-based system for supervisory control of rolling mills," added Dr Browne. Aluminium plates, used in the manufacture of aeroplanes, ships, etc, are produced by passing an initial aluminium slab through a giant 'mangle' (4-high Rolling stand) many times so that it is flattened into a thin final plate. A large number of factors may affect the quality of the final plate – the number of passes, plate temperature, plate width, and so forth – and as much background knowledge as possible is needed to ensure that the plates are near perfect, with no 'waves'. "With even the most advanced supervisory control system, not all plant knowledge is fully utilised," said Dr Browne. "However, our new system makes a much improved use of both the human knowledge of the plant and technological data. "It predicts and diagnoses quality problems before and during rolling and this leads to superior product shape and shape prediction, and improved explanations of product quality defects, coupled with suggested remedy actions. "Ultimately, the development of this knowledge-based system can improve the operational efficiency of plants manufacturing hot band aluminium plate by increasing productivity and yield." The EPSRC-funded research (following on from groundbreaking work by University of Leicester) involved the combining of Knowledge-Elicitation and Data-Mining techniques to develop the Fused Expert System. Knowledge-Elicitation involves establishing important facts and heuristics (rules of thumb) from plant experts, while Data-Mining is the process of analyzing data, often utilising advanced artificial intelligence techniques, in order to identify patterns or relationships. "The fusion of these two techniques produced an Expert System that successfully rolled aluminium plate without significant shape defects," said Dr Browne. "The methodology is transferable to all the other plate alloys and it is applicable to many other industrial problems. "The system could be used in many areas of industrial processing, such as drink cans or wallpaper and we hope that, with the aid of industrial sponsorship, we will be able to take the technology even further forward." end Media contacts: Dr Will Browne, Lecturer in Cybernetics, School of Systems Engineering, University of Reading T: 0118 378 6705 E: Craig Hillsley, Press Officer, University of Reading T: 0118 378 7388 E:

We use Javascript to improve your experience on, but it looks like yours is turned off. Everything will still work, but it is even more beautiful with Javascript in action. Find out more about why and how to turn it back on here.
We also use cookies to improve your time on the site, for more information please see our cookie policy.