Abstract
The safe decommissioning and dismantling of nuclear facilities requires a detailed and continuously updated understanding of the radiological environment. Traditionally, this process has relied on extensive measurement campaigns and conservative assumptions about radioactive source terms. While effective, such approaches are labour-intensive and may not adequately capture the dynamic evolution of radiological conditions as dismantling operations progress. Recent developments in artificial intelligence, machine learning, and digital twin technologies offer the potential to transform this process by enabling adaptive and predictive modelling of radiological risks in near real time.
This PhD project aims to develop an AI-driven dynamic source-term model that can be integrated within a digital twin framework of the dismantling environment. The objective is to enhance radiation protection by allowing continuous assessment and forecasting of radiological conditions as the work advances. The research will explore how data from gamma detection systems and operational parameters can be processed using AI and machine learning methods to estimate and predict changes in the radiological source term. By integrating these models into a digital twin, the system will provide real-time feedback that supports decision-making and improves compliance with the ALARA (As Low As Reasonably Achievable) principle.
The candidate will combine data-driven approaches with established physical models of radiation transport and decay, and will design data assimilation methods that dynamically update the source-term model as new measurements become available. The project will be conducted in collaboration with experts in radiation protection, digital twin technologies, and artificial intelligence, with validation performed using experimental or simulated datasets representative of decommissioning scenarios. Through this work, the PhD will contribute to the development of a new generation of intelligent tools for radiation protection, leading to safer, more efficient, and more transparent dismantling operations. The expected outcomes include a validated prototype of a dynamic source-term model integrated in a digital twin environment, along with methodological advances in coupling AI with radiological modelling and quantitative evidence of improved safety and operational efficiency.
Researcher(s)
Research team(s)
Funding
Project type(s)