Sustainable advanced manufacturing via machine learning-assisted exploitation of sensing and data infrastructure

AI, and data science in general, are underpinning research disciplines that already are very prominent in numerous manufacturing sectors, and are already achieving measurable impacts (e.g. in the high value, low volume sector in aerospace, automotive etc.) The applications supported via research in AI and data science in manufacturing
are typically around process modelling, monitoring, optimisation, control and decision making. These are typical engineering themes, now enhanced and supported via the exploitation of data. Frequent challenges that arise at the start of such initiatives are 1) what AI can do for a given process, and 2) how much and what kind of data do
we need? Pertinent to Machine Learning (ML) these questions cannot be answered in a trivial fashion. There are however mathematical and computational methods that can be used to establish the impact of data sparsity/availability on the effectiveness of various ML tools that can be subsequently used for process monitoring, forecasting etc. In this proposal we wish to explore feasibility studies, as a way of evaluating research possibilities and potential AI use in the foundation industries. The intention is to use available data retrospectively and demonstrate data utility, potential sensing needs, and ML forecasting capability. The overarching goal is to show that sustainability targets, as well as overall process performance, could be further enhanced via the use of appropriate data. Via this study, we also envisage identifying data challenges specific to each sector’s case studies, as well as directions for setting the future research agenda.

Dr. Iñaki Esnaola

University of Sheffield

esnola@sheffield.ac.uk