Magneto Hydrodynamics Real-time Detection on HT-7 Tokamak Device Based on RBP Neural Network

Abstract

The instability of Magneto Hydrodynamics (MHD) in tokamak plasma is a main factor in deciding high performance operation of the device. The occurrence of MHD instability will lead to deterioration of plasma confinement and even split of plasma discharge in severe instance, which can poke potential risk of damage to the device and its work staff. This paper presents a HT-7 MHD real-time detection system based on Radial Basis Probabilistic Neural Networks (RBPNN). The article firstly expands on measurement of MHD in HT-7 and corresponding character analysis of it. According to the signal frequency of MHD, RBFNN training samples can be constructed via mass data acquired through repeated discharges and thus completes the task of sample training. During the discharge, high speed data acquisition board DAQ2010 with double buffer is used to finish the job of real-time data acquisition while the trained RBPNN works spontaneously to process MHD signal. Repeated Tokamak discharges proved the effectiveness of the method described above.