Measurements of the vacuum-plasma response in EXTRAP T2R using generic closed-loop subspace system identification

A multibatch formulation of a multi-input multi-output closed-loop subspace system identification method is employed for the purpose of obtaining control-relevant models of the vacuum-plasma response in the magnetic confinement fusion experiment EXTRAP T2R. The accuracy of the estimate of the plant dynamics is estimated by computing bootstrap replication statistics of the dataset. It is seen that the thus identified models exhibit both predictive capabilities and physical spectral properties.

► Unstable plasma response safely measured using special signal processing techniques.
► Prediction-capable MIMO models obtained.
► Computational statistics employed to show physical content of these models.
► Multifold cross-validation applied for the supervised learning problem.