Conceptual design for real-time measurement of isotope ratio and RF electric field and in the scrape-off layer

We present a conceptual design for a real-time feedback control loop utilizing experimentally measured parameters from the Dynamic Stark Effect Low Hybrid Field (DSELF) diagnostic. DSELF is a high-resolution spectroscopic diagnostic that measures the polarized D_β/H_β line profile using optical emission spectroscopy. By properly modeling the line profile, we can extract information about the plasma isotope ratio and the lower hybrid (LH) wave RF electric field vector. Least squares analysis of the spectral data is computationally expensive and takes tenths of minutes for convergence due to the large number of free parameters required in the model. To overcome this limitation, we employ neural networks that provide the necessary throughput for real-time control with enough accuracy (uncertainty of 0.5% for the isotopic ratio and 0.09 kV/cm and 0.4 kV/cm for the field radial electric and RF poloidal, respectively). We investigate with DSELF the isotope ratio (D-H), which is an important parameter for ion cyclotron resonance heating (ICRH), and discuss diagnostic applicability to D-T ratio estimation. The other quantity inferred directly from the DSELF spectrum is the lower hybrid RF electric field, which is essential to quantify the propagation of LH waves through the edge of the plasma. That information can be used to optimize and control LH launchers via actuators (i.e., impurity powder dropper, supersonic molecular beam injection, pellets, etc.) in a feedback control loop.