Neural network evaluation of reflectometry density profiles for control purposes

Broadband reflectometry is a diagnostic that is able to measure the density profile with high spatial and temporal resolutions, therefore it can be us ed to improve the performance of advanced tokamak operation modes and to su pplement or correct the magnetics for plasma position control. To perform t hese tasks real-time processing is needed. Here we present a method that us es a neural network to make a fast evaluation of radial positions for selec ted density layers. Typical ASDEX Upgrade density profiles were used to gen erate the simulated network training and test sets. It is shown that the me thod has the potential to meet the tight timing requirements of control app lications with the required accuracy. The network is also able to provide a n accurate estimation of the position of density layers below the first den sity layer which is probed by an O-mode reflectometer, provided that it is trained with a realistic density profile model. (C) 1999 American Institute of Physics. [S0034-6748(99)61801-9].