Disruption prediction based on fusion feature extractor on J-TEXT

作     者：郑玮 薛凤鸣 陈忠勇 沈呈硕 艾鑫坤 钟昱 王能超 张明 丁永华 陈志鹏 杨州军 潘垣 Wei Zheng;Fengming Xue;Zhongyong Chen;Chengshuo Shen;Xinkun Ai;Yu Zhong;Nengchao Wang;Ming Zhang;Yonghua Ding;Zhipeng Chen;Zhoujun Yang;Yuan Pan

作者机构：International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma PhysicsState Key Laboratory of Advanced Electromagnetic Engineering and TechnologySchool of Electrical and Electronic EngineeringHuazhong University of Science and TechnologyWuhan 430074China Institute of Artificial IntelligenceHuazhong University of Science and TechnologyWuhan 430074China 

出 版 物：《Chinese Physics B》 (中国物理B（英文版）)

年 卷 期：2023年第32卷第7期

页      面：12-23页

核心收录：

学科分类：08[工学] 082701[工学-核能科学与工程] 0827[工学-核科学与技术] 

基　　金：Project supported by the National Key R&D Program of China (Grant No. 2022YFE03040004) the National Natural Science Foundation of China (Grant No. 51821005) 

主　　题：feature extractor disruption prediction deep learning tokamak diagnostics 

摘      要：Predicting disruptions across different tokamaks is necessary for next generation *** large-scale tokamaks can hardly tolerate disruptions at high performance discharge,which makes it difficult for current data-driven methods to obtain an acceptable result.A machine learning method capable of transferring a disruption prediction model trained on one tokamak to another is required to solve the *** key is a feature extractor which is able to extract common disruption precursor traces in tokamak diagnostic data,and can be easily transferred to other *** on the concerns above,this paper presents a deep feature extractor,namely,the fusion feature extractor(FFE),which is designed specifically for extracting disruption precursor features from common diagnostics on ***,an FFE-based disruption predictor on J-TEXT is *** feature extractor is aimed to extracting disruption-related precursors and is designed according to the precursors of disruption and their representations in common tokamak *** inductive bias on tokamak diagnostics data is *** paper presents the evolution of the neural network feature extractor and its comparison against general deep neural networks,as well as a physics-based feature extraction with a traditional machine learning *** demonstrate that the FFE may reach a similar effect with physics-guided manual feature extraction,and obtain a better result compared with other deep learning methods.