Abstract

Machine Learning is used in this paper for detecting anomalies in nuclear plant reactor cores. The proposed approach first generates large amounts of simulated data with different types of perturbations occurring at various locations in the core. This is achieved using the CORE SIM+ modelling framework, which generates these data in the frequency domain. State-of-the-art machine and deep learning models are then extended and used to successfully perform semantic segmentation of the core, classification and localisation of perturbations. Actual plant data are then considered, provided by two different reactors, including no labels about perturbation existence. A domain adaptation methodology is then developed, which uses self-supervised, or unsupervised learning, so as to align the simulated data with the actual plant data and detect perturbations, whilst classifying their type and estimating their location. Experimental studies illustrate the successful performance of the developed approach and extensions are described that indicate a great potential for further research.