Every year I have the opportunity to talk about my research at a symposium organized by the Swedish center for nuclear technologies (Svenskt Kärntekniskt Centrum). This year was the first time that we had a combined session with researches from ANItA (Academic-industrial Nuclear technology Initiative to Achieve a sustainable energy future). It was a nice experience to hear from other doctoral students, particularly those working on applied machine learning projects.
This year I spoke about the major research works which I had completed in the previous year, namely:
- The use of a recurrent neural network for predict fuel nuclide composition evolution. This work was recently submitted to the journal Annals of Nuclear Energy and is currently under review.
- The implementation of physics informed neural networks (PINNs) to improve model behavior for the prediction of Critical Heat Flux (CHF). This research was performed as a participant in the AI/ML benchmark series organized by OECD/NEA. More information can be found here.
The presentation slides can be found below. I had quite a lot of fun designing the slides this time (spending an unnecessarily long time thinking of the “two wolves” joke in the last four slides).