Intelligent recognition and location of the edge coherence mode in EAST

Abstract

Abstract In the study of multiple pedestal coherent modes in EAST, the high-precision intelligent recognition and location of the edge coherent modes (ECM) is a very meaningful work. In this study, a convolutional neural network (CNN) feature extraction model based on the attention mechanism was constructed to classify and localize the ECM in the EAST experiment. The classifier identified ECM on the test dataset with an accuracy of 0.970; the localizer detected the ECM with an average precision (AP) of 0.956. In addition, we found a strong correlation between the stored energy and current in plasma and the excitation of the ECM during EAST discharge; the triangularity may affect the relative amplitude of the ECM.