About the project
Conditional Generative Adversarial Networks (cGANs) have gained notoriety in the media for their ability to create so-called “deep fakes” but their power can also be harnessed to provide predictions of optimal engineering designs. Such techniques offer the potential to significantly reduce design times by providing engineers with an instant solution to a design problem.
Recent work by the University of Southampton has demonstrated that through a novel method of encoding information within the training data, a cGAN can be used to generate new designs which satisfy a desired performance criterion. This PhD will build upon this work and recent advances in 3D cGAN technology as well as octree and graph-based convolution neural networks to extend our capability to 3D dimensional problems with multiple objectives/constraints. Ultimately we aim to apply this novel generative approach to the design of gas turbine engine components and sub-systems.
This PhD is sponsored by Rolls-Royce Plc. and will be hosted within the Rolls-Royce University Technology Centre for Computational Engineering at the University of Southampton. The successful candidate will have opportunities to engage with Rolls-Royce engineers on several levels throughout the project, including a 3-month secondment to the company, which will offer a unique insight into the industrial realities of gas turbine engineering and maximise the wider impact of the research.
In addition to the basic tax free student stipend of £18,198 pa Rolls-Royce will provide a further tax-free stipend increment of £9,000 pa which brings the total stipend in line with their graduate recruitment schemes. The stipend will rise in subsequent years. Funding for travel to international conferences will be available.
