Aerospace Centre of Excellence Research in Computational Intelligence

Many engineering applications require the solution of global single or multi-objective optimisation problems, including mixed variables, multi-modal and non-differentiable quantities. From global trajectory optimisation to multidisciplinary aircraft and spacecraft design, from planning and scheduling for autonomous vehicles to the synthesis of robust controllers for airplanes or satellites, computational intelligence techniques have become an important – and in many cases inevitable – tool for tackling these kinds of problems, providing useful and non-intuitive solutions.

In particular the team is working in developing techniques and solving real-world engineering applications in the following fields:

  • Continuous and mixed integer optimisation: the best set of continuous and discrete parameters  is sought to minimise a perfomance measure. Methods for dealing with complex lanscapes such as multiple minima and irregularities in the model  can be used to solve the problem.
  • Combinatorial optimisation: the optimal solution from a finite set of objects is found. Routing and scheduling problems are an example. Enumerating all the different possibilities is usually intractable, space pruning techniques can be used to solve the problem.
  • Shape optimisation: the reference geometry is deformed in a continuous space to find the minimum of a cost function. Shape parametrisation techniques and multifidelity optimisation can be used to solve the problem.
  • Optimal control: the best control law for a system towards a predefined objective is sought. Direct and indirect methods can be used to solve the infinite dimensional problem.
  • Multiobjective optimisation:  the design space is explored towards the optimisation of two (or more) controversial objectives. The solution returned is a set of Pareto optimal solutions representing the best trade-off between the objectives.
  • Multifidelity optimisation: when computational intensive model are needed to assess performance or constraints of the problem, approximation techniques can come in hand to make the optimisation problem computationally tractable. Different levels of fidelity can be interfaced with the optimisation algorithm.
  • Multidisciplinary design optimisation: optimisation of a complex systems with multiple interacting disciplines. The system is optimised as a whole exploiting all the interactions between the different disciplines. Different problem formulations can be used to solve the multidisciplinary optimisation problem.´╗┐
  • Uncertainty quantification and propagation: development of intrusive and non-intrusive techniques 
  • Optimisation under uncertainty: treatment of aleatory and epistemic uncertainties with probability-based and evidence based techniques for robust optimisation 
  • Data-driven optimisation: modelling uncertainties of the optimisation problem from real life data 
  • Natural Language Processing, Machine Learning and Knowledge Management in support of space system design

Research & industrial projects

  • Multi Population Adaptive Inflationary Differential Evolution Algorithm (MP-AIDEA)
  • Multi Objective Hybrid Optimal Control
  • Evidence Based Robust Optimisation (EBRO)
  • Robust Data-Driven Combinatorial Optimisation
  • Uncertainty Treatment and OPtimisation In Aerospace Engineering (UTOPIAE)
  • Optimization of particle spatial distribution on elastic-plastic behavior of particle reinforced composites
  • Cloud-based shape optimisation in manufacturing
  • Rotojet Pump Pickup Tube Optimization
  • Uncertainty based Multidisciplinary Design Optimisation of Space Transportation Systems´╗┐
Read more about the Intelligent Computational Engineering Laboratory