Intelligent Systems, Decision and Control research group

The group also undertakes research into decision support systems for socio-economic, healthcare and manufacturing systems, control and optimization and smart materials.

Research themes

Intelligent Systems, Decision and Control, involves a diverse range of research, which includes the development of both core underlying theories and focussed application areas. Most of the research is funded from the research councils and similar bodies. A brief description of a selection of current research topics is given below.

Multi-Criteria Decision-Making

Our optimization research is strongly influenced by our experience in real-world applications. And, since the purpose of our research is problem-solving, our portfolio of optimization methods is not tied to any specific methodology. Applications experience covers a broad sweep of sectors such as engineering, medicine, pharmaceuticals and social issues.

For example, in manufacturing industries, such as the aerospace and automotive sectors, design typically requires consideration of multiple competing performance criteria, or objectives, for which a solution is sought that is robust to underlying assumptions and uncertainties. In such a case, our research seeks to construct practical optimization workflows that integrate a model-based systems engineering design approach with a multi-criteria decision making methodology in order to realize robust design solutions.

Research interests include: (i) how to handle many (>3) competing objectives; (ii) how to manage exogenous and endogenous uncertainty; (iii) methods to address the cognitive challenge of comparing and down-selecting solutions in high-dimensional spaces; (iv) methods to facilitate joint decisions by multiple decision makers; and (v) how to enable a network of decision support and optimization nodes to operate asynchronously.

Modelling and Optimisation for Manufacturing

The group has created a 'right-first-time' modelling and optimisation framework for manufacturing, based on the concept of 'granular computing'. The framework has been used in the manufacturing of fuselage structure, for Boeing, Embraer and Eclipse, and other products.

We developed a new data-mining and modelling framework using the theory of granular computing and the evident impact of this framework includes the model-based evaluation of Friction Stir Welding as as well as fault detection in steel manufacturing.

Modelling of Systems Behaviour for Monitoring and Prediction

Characterising behaviour of systems when the underlying physical principles are either uncertain or too complex to derive and when the observations are only partially complete, is challenging. By building mathematical and statistical modelling based on observations, we aim to monitor changes to system behaviours and to make predictions of how they will evolve in the future.

We have developed novel technologies for monitoring the health of systems and applied them to aircraft power systems. The group’s breakthrough research introduced the first quantitative modelling framework of conflict dynamics using data from Wikileaks on the Afghan War which won the prestigious 2012 PNAS Cozzarelli Prize.

Modelling of Social Systems

The group makes a leading contribution to interdisciplinary research on health policy appraisal, including participating in the development of the Sheffield Alcohol Policy Model, one of the most well-known and controversial mathematical models developed in recent years. The model and its results have made a major contribution to the international policy debate on alcohol pricing.

These seminal modelling methods are being applied and developed in further research grants, including development of agent-based modelling approaches to explain and predict alcohol consumption behaviours in the British population.

Control Design and Predictive Control

The group has specific expertise in predictive control design techniques and has created concepts which are now broadly adopted across the research community such as: (i) the use of closed-loop predictions to improve problem conditioning and stability; (ii) systematic constraint tightening to ensure safe disturbance handling and (iii) mechanisms for reducing optimisation complexity. Current work is focused on effective tailoring for specific industrial scenarios.

Collaborations

The Intelligent Systems, Decision and Control research group collaborate widely with industrial partners and other universities:

• Rolls-Royce University Technology Centre (UTC)
• TATA-Steel Europe
• Jaguar Land Rover
• Rail Innovation and Technology Centre (RITC)
• Airbus