Prediction for Tokamak High-power Pulse Load based on Elman Neural Network

Abstract

A dynamic load prediction model of the Elman neural network is suggested to increase the reliability of the tokamak high-power pulse prediction. The method enhances the network’s capacity to process dynamic input by employing a dynamic back-propagation algorithm. A comparison of the simulation prediction results and measured data, depending on the measured data of the EAST (the experimental advanced superconducting tokamak) power bus, shows that the prediction results of Elman are more accurate than the traditional BP neural network, which can describe the impact of the fusion device’s pulsed load on the power grid better and increase the precision of stability analysis.