The funding, from the UK Government’s Engineering and Physical Sciences Research Council (EPSRC), will help Imperial College London experts to set up and run the programme, known as PREMIERE, over five years.
The researchers expect the programme , which will be based on machine learning-powered computer simulations, to improve supply chain design, decision-making, safety management, and help reduce carbon emissions.
Principal Investigator Professor Omar Matar , of Imperial’s Department of Chemical Engineering , said: “This funding will help scientists and engineers train algorithms to solve energy, manufacturing, and healthcare problems based on artificial intelligence and machine learning.”
We hope the project will result in algorithms that increase the efficiency of manufacturing processes, improved safety management, reduce carbon footprints, and improve patient-specific medicine. Professor Omar Matar Department of Chemical Engineering
PREMIERE stands for PREdictive Modelling with QuantIfication of UncERtainty for MultiphasE Systems. It will be run by a multidisciplinary collaboration including Imperial, the University of Birmingham , University of Cambridge , UCL , and the Alan Turing Institute.
The team will focus on three main sectors of interest: manufacturing, energy, and healthcare. Their aim is to create the next generation of models for multiphase flow systems - systems that deal with flow of gas, liquid, and, potentially, solids flowing simultaneously in pipes, channels, and reactors - to enhance productivity and efficiency across these sectors.
The PREMIERE project seeks to better understand the behaviour of these flow patterns and use machine learning and computer simulations to predict how these patterns can be influenced for the better - whether the flow is in the body or in a chemical plant.
Professor Matar said: “Through PREMIERE we will develop new, intelligent ways to manage uncertainty, minimise risk, boost efficiency, reduce carbon emissions, and improve patient care.
“Multiphase flow systems are something we all experience daily, but many of us are unaware of them. Despite their importance, many areas of these flows have yet to be studied in detail - which is why PREMIERE is important.”
ManufacturingManufacturing plays a major role in the UK and global economy, but unpredictability in demand, availability of raw materials, and variations in consumer preferences can cause uncertainty in the industry. To manage these risks, manufacturers need their processes to be as efficient, sustainable, and robust as possible.
Through PREMIERE, machine learning algorithms could be used to predict, and help people plan for, supply chain interruptions. For example, a raw materials-producing country may be unable to meet demand due to economic or geopolitical upheaval. An artificially intelligent algorithm would constantly keep track of how this might affect supply chains, warn about risk of disruptions, and suggest viable alternatives from which industrialists can choose.
Professor Matar said: “Machine learning algorithms could help the manufacturing industry work with evolving economic, environmental, and geo-political landscapes.”
PREMIERE will look to develop predictive frameworks through the combination and tight integration of data and mechanistic models Professor Omar Matar Department of Chemical Engineering
The PREMIERE project will help to deploy energy systems that will make a significant contribution to generating low-carbon power. This will support the UK in meeting its net-zero emissions pledge by 2050.
Smart algorithms could help oil-and-gas companies to reduce emissions by improving efficiency and productivity. Underwater pipes are often used to transport oil at low sea temperatures. Smart algorithms could monitor and predict how these temperatures could lead to pipe blockage, prompting engineers to release the right chemicals into the oil, saving energy costs from trying to unblock pipes.
Professor Matar said: “Flows comprising oil, water, and air, are complex and poorly understood. PREMIERE will look to develop predictive frameworks through the combination and tight integration of data and mechanistic models.”
HealthcareData-driven algorithms could help to inform personalised patient care and predict specific conditions that are difficult to diagnose correctly.
For example, Imperial will work with clinicians from University of Birmingham’s Surgical Reconstruction and Microbiology Research Centre to collect data on acute compartment syndrome. This is a life-threatening condition often seen in patients from road traffic accidents, and is difficult to diagnose accurately. A wrong diagnosis can lead to amputation or death.
Through PREMIERE, smart algorithms that interpret patient data could be used to diagnose patients based on individual symptoms and those of previously diagnosed patients, potentially leading to more accurate diagnoses.
Professor Matar said: “Enabling early diagnosis in the presence of uncertainty requires improved understanding of how symptoms relate to correct diagnoses. Data-driven personalised therapies could help address cases like this and bring a wealth of new personalised medicine applications.”
The researchers behind PREMIERE expect their algorithms to be flexible enough to offer a whole range of solutions, as part of the same framework, depending on the needs of their industrial and medical partners: from approximate solutions delivered in real-time, to highly-accurate ones that require use of powerful, high-performance computing.
PREMIERE will be funded by the EPRSC, which is part of UK Research and Innovation. This will support a team of PhD and postdoctoral research associates, as well as research software engineers across the four partner universities working closely with industrial and medical partners.
Professor Matar said: “We hope the project will result in algorithms that increase the efficiency of manufacturing processes, improved safety management, reduce carbon footprints, and improve patient-specific medicine.” Photos and graphics subject to third party copyright used with permission or © Imperial College London.
Communications and Public Affairs