Jiuying Li

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 2.5 × 10 9 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.

A module with self-consistent evolution of driven current is developed and coupled with the resistive-MHD equations in the three-dimensional, toroidal, and nonlinear simulation code (CLT). The driven current equation is solved with a second-order accuracy symmetric scheme, which exhibits good conservation properties. With the new module, we find that the driven current can self-consistently concentrate inside the magnetic island when the parallel diffusion of the driven current is sufficiently large. The efficiency of the driven current on tearing mode suppression will then be much higher than those with stationary distributions. With the new module, the influence of electron cyclotron current drive (ECCD) on the nonlinear evolution of the 2/1 double tearing modes (DTMs) is investigated. When co-ECCD deposits on the outer resonant surface, the local magnetic shear is reduced, and the growth rates of the DTMs decrease; if ctr-ECCD deposits on the outer resonant surface, the local magnetic shear increases, and the DTMs become more unstable. However, things will be different if ECCD deposits on the inner resonant surface since the local magnetic shear is negative. The co-ECCD deposited on the inner resonant surface increases the negative shear and then promotes the growth of the DTMs; while the ctr-ECCD suppresses the DTMs. It is also found that the off-axis and central pressure crashes associated with the 2/1 DTMs can be converted to each other by properly depositing the driven current. To convert a central crash to an off-axis crash, the co-ECCD should be deposited on the outer resonant surface, or the ctr-ECCD deposited on the inner resonant surface. While, the co-ECCD should be deposited on the inner rational surface, or the ctr-ECCD deposited on the outer rational surface to convert an off-axis crash to a central crash. The co- or ctr-ECCD should be larger than a threshold for such transitions, and the threshold value is mainly determined by the location of the inner resonant surface.

The pressure crashes observed in shot No. 71326 in the Experimental Advanced Superconducting Tokamak are numerically investigated with the three-dimensional, toroidal, and full resistive-magnetohydrodynamics code (CLT). Based on the experimental observations, the pressure crash is caused by the nonlinear evolution of the m/n = 2/1 double tearing mode (DTM), where n and m are the toroidal and poloidal mode numbers, respectively. However, we find that the m/n = 2/1 DTM is stable based on the safety factor (q) profile from the equilibrium code EFIT, which indicates that the original q profile is somewhat inconsistent with the actual profile due to q measurement uncertainty. Since there is no motional Stark effect diagnostic for this shot, the local information of the magnetic field is missing, which leads to the largest contribution to the discrepancy. If other information is perfectly known and the q profile is the only uncertainty, then we could provide some information for the EFIT reconstruction by comparing our simulation results with electron cyclotron emission signals to constrain the uncertainty of the q profile to a much smaller region. The influence of plasma rotation and the two-fluids effect is also discussed.