Pursuing the smart factory

Materials World magazine
10 Dec 2018

Given Crown’s specialities, is your project focused on factories working with aluminium, or extended beyond? 

The collaboration with Crown is part of a larger EPSRC-funded Fellowship research project. The overarching aim of my project is to develop novel computational resources and methodologies, based on Machine Learning, which will enable to predict anomalies such as failure conditions or machine faults in smart factories. Machine Learning (ML) is a well-established branch of Artificial Intelligence (AI) that entails the ability of computers to learn based on large volume of data or experience and is already applied to several domains such as finance, data security and marketing. My project vision is to develop resources that will facilitate the adoption of ML technologies in factory environments, harnessing the potential to learn and make predictions from large volume of data that are routinely collected.

The project scope extends beyond aluminium processing industry. For instance, I aim to apply the same techniques to steel making processes. The research is also part of a larger research institute, IMPACT (Innovative Materials, Processing and Numerical Technologies), which provides a highly specialised and dynamic research environment for collaboration of industry and academia in the field of computational engineering modelling and materials science, applied for instance to manufacturing engineering. IMPACT will work on fundamental research, whereas Swansea University’s new Factory of the Future seeks to exploit these through accelerated innovation in close collaboration with industry.


Will the focus be kept within packaging production, or more widespread?

I aim to develop technologies that can have a widespread application to different manufacturing sectors, beyond packaging production. By working with companies in different domains, my vision is to be able to transfer knowledge across domains, and, ultimately through my research, help UK manufacturers to increase the uptake of data analytics and ML technologies. This will help companies to develop smart factories that can adapt to changes more rapidly and can self-optimise. Whilst I aim to develop generic methodologies and tools, I am also keen to demonstrate and fully evaluate the use of these technologies and resources in real application domains. This is why I have joined up with industrial partners such as Crown to pioneer the application of predictive data driven technologies to their packaging production to drive optimal performance.


Can you provide more information on the semi-autonomous processes you’ll investigate?

Despite all the hype around the next generation of fully autonomous and intelligent factories, also called Industry 4.0, I believe we are still far from achieving this vision. As a first step towards development of fully autonomous and intelligent systems, new technologies such as ML offer great potential, enabling organisations to analyse complex data and develop predictive models that can help them to make data-driven decisions about their operations. For instance, ML technologies can be used to detect patterns of failures and give suggestions on the best course of action for recovery. In the short term, the insights developed can be used to augment and support human decision-making capabilities. For this reason, my research does not only focus on the development of novel algorithms and computational models but also seeks to provide a better understanding of the interaction between human and machine intelligence in real factory environments. As part of my research, I aim to develop demonstrators of the technology that will enable us to gain a better understanding of its potential as well as limitations and give recommendations for real world implementations. The two main applications will be in metal packaging and steel production, but I aim to extend the application domain to other industries.


You mention in the Swansea University press release that UK manufacturers and suppliers are falling behind international competitors. Is this specifically referring to its smart factories success, and why has this become the case, given the proliferation of the UK’s wider digital industries?

In term of digital industries, I agree that the UK is in a position of strength and has a strong ambition to be at the forefront of AI and data revolution, as highlighted in the Industrial Strategy. In the press release, I am referring to the productivity challenge the UK is facing. Although certain sectors are performing well (i.e. automotive and aerospace), as a nation, productivity levels are lower than our EU competitors’ and other international countries (USA, Japan, Canada). Workers in the UK produce approximately the same amount in five days as the average G7 worker achieves in four days. Against this backdrop, the recently published Made Smarter review, commissioned by the government, has specified industrial digital technologies as key to driving productivity in the UK Manufacturing Sector and has set out a vision for the UK to become leader in industrial digitalisation. My research is aligned to the objectives of the Made Smarter review by contributing to increasing adoption of AI and ML technologies in smart factories.