Fusion Futures: Uncertainty Quantification for Plasma Control Systems

Description

The project objective is to extend real-time plasma control frameworks with uncertainty quantification (UQ) to increase resilience. Plasma control systems must operate reliably. They have tight real-time performance constraints and must work across a range of parameter space in which adapting effectively to the response of the highly non-linear plasma is critical. Modern approaches to control problems recognize the importance of quantifying uncertainty and applying systematic treatment to optimise results. The project seeks to apply state of the art thinking from the field of uncertainty quantification to build new capabilities in plasma control software frameworks. In general, there is scope to make control systems more resilient and to better characterize and control operational risks by applying UQ methods in plasma control system. Example objectives could include studies of the variable performance of control systems as a function of actuator response variability, either static, or time varying (the latter including short term and long term drift). Approaches could be first principles based on parametric knowledge of behaviours from models, and/or empirical inferring statistical variability from data. Another direction of interest is applying UQ to control of systems with uncertain dynamics and building robust solutions even in the absence of detailed models. UKAEA require the provision of staff with strong knowledge and experience of the theory and techniques in modern uncertainty quantification to work in close cooperation with UKAEA research and development staff for a period running from contract award until March 31 2025.