Resilient and robust infrastructure
Major infrastructure systems such as railways, power plants and supply chains are vital to our way of life. Increasingly these are becoming intense generators of data, and their safe operation is dependent on data science.
We are applying modern data-centric engineering techniques to perform predictive modelling of London Underground trains, designing sensor operation for water distribution networks, and analysing large-scale usage of roads across different modes of transport.
These projects and others in development are all contributing to operating more resilient infrastructure systems, robust against shocks or unexpected incidents and preventing future ‘black sky’ events.
This challenge area is led by Professor Julie McCann, Professor of Computer Systems at Imperial College London.
Sustainable infrastructures
Combining reliable sensing technologies with online data analytics to improve real-time sustainability and resilience of infrastructures
Monitoring of complex systems
Understanding and anticipating the impact of rare and high-consequence events in complex engineering systems such as aircraft, power stations or cyber networks, is critical for safety and security in the modern world.
We are using algorithms and statistical techniques to perform health monitoring on vital railway bridges, operate predictive maintenance on gas turbines, and predict rare events in transport systems. We are also modelling extreme events in cyber networks to detect malicious attacks on critical infrastructure.
These projects allow us to operate old and new infrastructure in smarter, safer ways, making best use of data to preserve the operation of vital systems.
This challenge area is led by Professor Emma McCoy, Professor of Statistics at Imperial College London.
Instrumented infrastructure
Developing new statistical methods for sensor data to improve how infrastructure is delivered, maintained and controlled
Extreme event prediction and monitoring
Developing statistical methodology for dealing with extreme events in cyber and transportation systems involving multiple different variables
Spatio-temporal causal inference
Developing methodology for identifying causal connections in settings that have relationships in space and time, such as urban traffic movement
Data-driven engineering design under uncertainty
Sensors and monitoring equipment are being increasingly employed in engineering applications, from complex computer networks to detecting pedestrian traffic in urban environments. The data being generated presents enormous opportunities to transform both system design and control.
However, to take full advantage of modern technology we need systems that are designed for data, and designed by data. The placement of sensors, the operation of autonomous systems and the way humans and machines interact must all be informed by data-centric engineering.
This challenge area will address fundamental questions of optimal data collection and design optimisation in uncertain environments.
This challenge area is led by Professor Jennifer Whyte, Laing O’Rourke / Royal Academy of Engineering Professor of Systems Integration at Imperial College London.
Retrofit design in the built environment
Exploring decision making processes common in construction and building management, to evaluate potential opportunities for retrofitting in the built environment
Data-driven design assurance
Using text mining and natural language processing to provide insight into design assurance practices; the process of assuring the right job is done the right way
Mathematical foundations
Rigorous theoretical research and practices are essential for delivering reliable and robust data science solutions within engineering applications.
The objective of the mathematical foundations theme is to facilitate and develop links between the theoretical researchers involved in the programme and the applied projects pursued within it. The aim is to to build a bridge between theory and applications across the programme.
Theoretical research is being conducted in areas including probability theory and mathematical statistics, for translation and application to data science and engineering.
This challenge is led by Professor Alex Mijatovic, Chair in Statistics at the University of Warwick.
Computational statistical inference for engineering and security (CoSInES)
Creating a step change in the use of statistical methodology, motivated by challenges in modelling, computation, and statistical algorithms
Estimating system health from streaming sensor data
Developing methods to estimate the health of engineering systems from real-time sensor data, to make predictions about the health in the future
Free energy landscapes for big data
Deriving efficient algorithms by understanding the natural geometric structure of data landscapes arising in machine learning
Human action recognition
Developing a framework to interpret complex multi-dimensional data streams of human actions, by utilising ‘path signatures’ and deep learning
Inverse problems
Developing statistical tools for problems in applications such as medical imaging and industrial safety, where unknown parameters are determined from observations and measurements
Mean-field games and interacting particle systems
Using models of strategic decision making in very large populations to study Smart Grids and statistical sampling
Education
It is vital that future generations of engineers appreciate the utility of data and understand how to correctly and astutely deploy modern data analysis techniques.
The programme for data-centric engineering aims to actively promote and develop education in data science. This includes dissemination of the programme’s research outputs among universities and industrial partners.
The education strategy of the programme is currently being developed, and we welcome input from interested parties.
The education theme is led by Professor Omar Matar, Vice-Dean (Education), Faculty of Engineering at Imperial College London.