A model-based approach for real-time estimation and control of the electron density profile: application of the upgraded RAPDENS code to TCV and ASDEX Upgrade

Future fusion reactors will have limited diagnostics coverage, and therefore sensors for feedback control. To augment the control system plasma monitoring, a model-based method is presented, based on the Extended Kalman Filter (EKF).

The RAPDENS code adopts an EKF to reconstruct the real-time electron density profile of tokamak plasmas, by combining diagnostic signals with a control-oriented predictive model. This approach provides an estimate of the density state while rejecting diagnostics faults, thus increasing control system reliability.

In this work, a novel boundary condition at the plasma separatrix is tested and validated with AUG and TCV offline data. The separatrix density, estimated with empirical formulas, scrape-off layer reduced models or via the EKF, provides an enhanced reconstruction of the density profile at the edge and reduces the tuning requirements of the predictive model.

The code has been integrated into the TCV control system, and its deployment on AUG is in progress. For TCV, it is adopted as a multi-rate observer, employing high spatial accurate Thomson scattering and high temporally resolved far infrared interferometer signals. Its use as a routine tool for a wide variety of physics experiments has been demonstrated, such as exhaust studies or high-density H-mode discharges. The improved density estimator has been validated in challenging experimental conditions, e.g. ECRH density pump-out, NBI core fueling and limited diagnostics coverage.