Physics-guided attention-aware convolutional neural networks for identification of magnetic islands in the tearing mode on EAST tokamak

Abstract The tearing mode (TM), a large-scale magnetohydrodynamic instability in tokamak, typically disrupts the equilibrium magnetic surfaces, leads to the formation of magnetic islands, reduces core electron temperature and density, thus resulting in significant energy losses and may even cause discharge termination. This process is unacceptable for International Thermonuclear Experimental Reactor (ITER). Therefore, the accurate identification of a magnetic island in real time is crucial for the effective control of the TM in ITER in the future. In this study, based on the characteristics induced by TMs, an attention-aware convolutional neural network (AA-CNN) is proposed to identify the presence of magnetic islands in TM discharge utilizing the data from ECE diagnostics in the EAST tokamak. A total of 11 ECE channels covering the range of core is used in the TM dataset, which includes <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mn>2.5</mml:mn> <mml:mo>×</mml:mo> <mml:mrow> <mml:msup> <mml:mn>10</mml:mn> <mml:mn>9</mml:mn> </mml:msup> </mml:mrow> </mml:mrow> </mml:math> data collected from 68 shots from 2016 to 2021 years. We split the dataset into training, validation, and test sets (66.5%, 5.7%, and 27.8%), respectively. An attention mechanism is designed to couple with the convolutional neural networks (CNNs) to improve the capability of feature extraction of signals. During the model training process, we utilized adaptive learning rate adjustment and early stopping mechanisms to optimize performance of AA-CNN. The model results show that a classification accuracy of 91.96% is achieved in TM identification. Compared to CNN without attention-aware, the AA-CNNs demonstrate great performance across accuracy, recall metrics, and F1 score. By leveraging the deep learning model, which incorporates a physical understanding of the tearing process to identify TM behaviors, the combination of physical mechanisms and deep learning is emphasized, significantly laying an important foundation for the future intelligent control of TM dynamics.