Artificial neural networks are trained to forecast the plasma disruption in HL-2A tokamak. Optimized network architecture is obtained. Saliency analysis is made to assess the relative importance of different diagnosti...Artificial neural networks are trained to forecast the plasma disruption in HL-2A tokamak. Optimized network architecture is obtained. Saliency analysis is made to assess the relative importance of different diagnostic signals as network input. The trained networks can successfully detect the disruptive pulses of HL-2A tokamak. The results obtained show the possibility of developing a neural network predictor that intervenes well in advance for avoiding plasma disruption or mitigating its effects.展开更多
基金Project supported by the National Natural Science Foundations of China （Grant No 10775040） and partially by JSPS-CAS Core University Program on Plasma and Nuclear Fusion.Acknowledgments The authors take this opportunity to express their sincere thanks to Q. D. Gao for his continuing encouragement and support. They gratefully acknowledge Y. Liu, B. B. Feng and F. Z. Li for fruitful discussions. Finally, the authors thank the entire HL-2A team for supplying the experimental data.
文摘Artificial neural networks are trained to forecast the plasma disruption in HL-2A tokamak. Optimized network architecture is obtained. Saliency analysis is made to assess the relative importance of different diagnostic signals as network input. The trained networks can successfully detect the disruptive pulses of HL-2A tokamak. The results obtained show the possibility of developing a neural network predictor that intervenes well in advance for avoiding plasma disruption or mitigating its effects.
