Research Of The Plasma Disruption Prediction And Runaway Electron Beams Control On J-TEXT

Tokamak is believed to be the most promising reactor for a controlled fusion.However,disruption is one of the major destabilizing factors that restrict the normal operation of large tokamak in the future.When a disruption occurs,a large amount of energy can be released in a moment.The event consists of two processes: thermal quench(TQ)and current quench(CQ).The damage caused by the disruption to the device is reflected in the following aspects:(1)plasma-conducted thermal loading of the plasma facing components during the thermal quench,(2)J×B forces from poloidal halo and eddy currents during the current quench,and(3)the conversion of toroidal plasma current into energetic runaway electrons(REs)that are eventually stopped by the first wall.In smaller machines,a disruption may be tolerated,but in a reactor-sized machine,a major disruption could damage the walls and other internal structures,require serious repairs,and cause severe setbacks in its operation.Therefore,it is necessary to take measures to avoid or mitigate the effect of disruption,and a set of feasible disruption prediction system is the key to effective implementation of these measures.This paper analyzes the main causes and possible physical processes of density limit disruption.From the perspective of literature research,it is believed that density limit disruption has a close relationship with the evolution of temperature profile and radiation profile.On the basis of previous researches,two sets of Back-Propagation neural network disruption prediction systems are established.A set of prediction system only choose the diagnostic signals in plasma core that have a close relationship with density limit as inputs to the neural network.In another system,in addition to the above signals,the signals that can reflect the evolution of the plasma density profile,temperature profile and radiation profile are also included.By comparing the prediction results of these two systems found that by adding more information in different positions of plasma to reflect the evolution of plasma parameters in the whole space can effectively improve the prediction effect of neural network.The purpose of the disruption prediction is in order to take the avoidance and mitigation measures of disruption in time.Whereas,once any part of the plasma disruption prediction or mitigation system does not work,there will be serious damage to the vacuum vessel and the plasma-facing-component of tokamak,especially the damage caused by runaway electrons.Recent experiments in the J-TEXT tokamak have demonstrated active control of the relativistic runaway beam position following a disruption.The aim is to maintain a runaway beam at a safe position within the vacuum vessel while gradually dissipating its energy,thereby reducing the potential threat that the beam poses to the in-vessel components of the tokamak.There are two active control schemes used: in the first scheme,under the condition that only horizontal displacement control system is added,the length of runaway plateau has been sustained about 30 ms longer than for the case that no active control has been applied;in the second scheme,where both the horizontal displacement control system and magnetic energy transfer system are put into use at the same time,the length of runaway plateau has been sustained 30 ms longer than horizontal displacement control system applied only.Owing to MET coils have the characteristics of transfer the poloidal magnetic energy of REs out of the vacuum vessel in the period of the current of runaway beams decreases,reducing the loop voltage of runaway electrons during the runaway current plateau,and decreasing the current of runaway electrons when enhanced wall interaction occurred.All of these are in favor of reducing the kinetic energy of runaway electrons,thereby reducing the damage to device.In the future experiments in J-TEXT,more attention should be paid to study on this control strategy.