Application of Delta-bar-Delta Rules Trained Back-Propagation Neural Networks in Nuclear Fusion Pattern Recognition

On the basis of controlled nuclear fusion equipment HT-7 superconductive tokamak´s detection data, this paper reports on an approach of nuclear fusion magneto hydrodynamics(MHD) pattern recognition by using artificial neural network and back-propagation(BP) neural network with delta-bar-delta rules which can monitor the system characteristics and recognize the MHD pattern precisely. The HT-7 nuclear fusion plasma´s electric current, pressure distribution and magnetic field etc shift the system status, and some of which make further efforts to cause the split of plasma which will result in disasters. MHD pattern is one of the most dangerous circumstances. So the recognition of MHD pattern becomes the most significant task. The experimental evidence strongly suggests this approach has obtained a favorable constriction rate and discrimination precision.