Development of Data Assimilation System for Temperature and Density Control in Helical Plasmas

抄録

We develop a real-time adaptive predictive control system based on data assimilation (DA) for the temperature and density of helical fusion plasmas. The DA-based control approach enables the harmonious integration of measurement, heating, fueling, and simulation and can provide a flexible platform for adaptive model predictive control. The core part of the control system, ASTI, is built upon the integrated simulation code TASK3D and a data assimilation framework DACS. DACS integrates adaptation of the predictive model (digital twin) to the actual system using real-time measurements and control estimation that is robust against model and observation uncertainties. We perform numerical experiments using ASTI to control the electron temperature profile and density of a virtual plasma generated by TASK3D. The results demonstrate that ASTI can effectively drive the virtual plasma state toward the target state while bridging the gap between the digital twin and the virtual plasma. Furthermore, the numerical experiments clarify the effects of hyperparameters in the DA-based control approach on control performance.