Modelling and Control of Flexible Structures Interacting with Fluids (ModConFlex)
Lead Research Organisation:
Imperial College London
Department Name: Aeronautics
Abstract
The ModConFlex consortium comprises a group of 10 academics and 4 senior researchers in industry (ORE Catapult) with expertise in control theory, artificial intelligence, complex dynamical systems, distributed parameter systems, fluid dynamics, aeroelasticity, power electronics, power systems, swimming theory and marine engineering. Our aim is to train the next generation of researchers on the modelling and control of flexible structures interacting with fluids (water and air), contributing to the latest advances in control theory, artificial intelligence and energy-based modelling. Our main applications are in the control of floating wind turbines (the prime renewable energy source of the future), and in the control of highly flexible aircraft, aircraft with very high aspect ratio. Our research plans are organized into three scientific work packages, which cover mathematical systems theory (modelling and model reduction, boundary control systems, port-Hamiltonian systems, exact beam theory), relevant aspects of control theory (internal model controllers with anti-windup, nonlinear model predictive control, robust control), reinforcement learning, aeroelasticity, stochastic algorithms. We believe that science and technology in Europe will greatly benefit from this research, and from the education and knowledge that we will impart to a new generation of researchers. Key strengths of this consortium include a research environment that brings together mathematicians and engineers to provide the project's young researchers with a unique training environment, and a network of associated industrial partners that will allow all the young researchers to participate in industrial secondments. We have the critical mass to cover all aspects of training, and we have an excellent track record of past collaboration and of training young researchers.
Publications
| Description | At the heart of the research questions of this project is the aim to find simple descriptions of complex dynamical systems, particularly those with aerospace applications. The challenge in aerospace is that the dynamical processes of interest, such as the response of an aircraft's wing to a gust of air, can be highly nonlinear. This makes predicting and controlling their behaviour-which is essential for safe and efficient flight-extremely challenging. To try to overcome this problem, we are exploiting AI-based data-driven modelling approaches to extract simple models from high-dimensional data sets in aerospace. Our research so far has found that it hard to use standard methods off-the-shelf, but that improved performance can be achieved by embedding knowledge of the underling physics into the model identification process. The first key finding of our research is this discovery that embedding knowledge of the deformed shape of an aircraft's wing at different fixed flight conditions is very useful for identifying good nonlinear models for a wing's dynamic motion. We found that models created using the new technique outperform state of the art AI-based modelling approaches on a challenging benchmark test-case of a flexible aircraft wing, known as the "Pazy Wing". Preliminary results in this direction were published at the recent American Institute of Aeronautics and Astronautics (AIAA) Science and Technology Forum in Orlando, Florida, in January 2025. The second key finding of our research so far is that we have created a new optimization method, called the Recursive Nonlinear Mode Method. This creates simple models of nonlinear structural mechanics systems, such as those describing the deformation of an aircraft's wing or the blade of a wind turbine. We have also developed a computer-assisted method of proving rigorous bounds on nonlinear energy transfer between different types of motion in nonlinear structural mechanical systems. We published these findings at the recent Institute of Electrical and Electronics Engineers (IEEE) Conference on Decision and Control (CDC) meeting in Milan, Italy, in December 2024. From the perspective of AI-based modelling, our research is important since it has highlighted a class of problems in aerospace that cannot be easily solved with standard AI tools. Our aim for the remainder of the project is to build on our successful approach to creating data-driven modelling and control methods which are tailored to the specific mathematical properties of the complex dynamical systems (i.e. the coupled motion of fluids and flexible structures) we are studying. |
| Exploitation Route | The planned outcome of our research is to create a novel suite of data-driven modelling and control methods that can improve the design of future aircraft. We expect that both academic and non-academic users may benefit from our research, due to the immediate need to decarbonise aviation. One way to help achieve this is to improve the fuel-efficiency of current and future aircraft. It is accepted that a good way to achieve this is by employing long, slender, wings which are both aerodynamically-efficient and also reduce the aircraft's weight and fuel consumption. However, lighter and longer wings must deform more than those of traditional aircraft, which implies that the motion of future aircraft will be more complex to predict. New advances in modelling and control theory that this project aims to deliver are therefore needed to let engineers find the best possible designs for safe, efficient, and low-carbon future flight. |
| Sectors | Aerospace Defence and Marine Transport |
| Description | 2-day training course to post-graduate researchers on model predictive control |
| Geographic Reach | Europe |
| Policy Influence Type | Influenced training of practitioners or researchers |
| Impact | This was an academic training course seeking to develop the skills of post-graduate students. |
| Description | Attendance of team to Training Workshop of EU MSCA ITN Training School, University of Twente, Netherlands, November 2023 |
| Organisation | University of Twente |
| Country | Netherlands |
| Sector | Academic/University |
| PI Contribution | Dr Wynn and the Reserach Assistant/PhD student employed on this grant, Mr Mario Sinani, attended a project meeting and training workshop of the EU ITN MSCA Doctoral Training Network "(ModConFlex) Modelling and Control of Flexible Structures and Fluid Structure Interactions". This grant is EPSRC matched funding for our participation in this training network. |
| Collaborator Contribution | Dr Wynn represented Imperial College London's role as an associated partner in this project. He taught a 2-day training school to all the PhD students from the European Training Network. Mario Sinani attended a week-long training school. |
| Impact | This is an ongoing EU Doctoral Training Network. Outcomes (publications) and collaborations (via secondments) are expected over the next four years of the project's duration. |
| Start Year | 2023 |
| Description | Presentation on control of high-aspect-ratio wing aircraft given to industry-academia event. |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | The PI, Dr Wynn, presented research at the ""AR20+: Workshop on High Aspect Ratio Wing Technologies" workshop at Imperial College London. This was attended by approximately 100 participants from the aviation industry (e.g. AIRBUS), governmental aviation research centers (e.g. DLR (Germany), ATI (UK)) and an international academic audience. Dr Wynn presented new research related to the funded project and led a discussion group on the future vision of control technologies for future aircraft. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://www.imperial.ac.uk/events/165806/ar20/ |