Abstract:In this study, we develop five deep learning models, a Convolutional Neural Network (CNN) model, a CNN model with Squeeze-and-Excitation Attention(CNN-SE), a CNN model with Convolutional Block Attention Module (CNN-CBAM), a CNN model with Efficient Channel Attention (CNN-ECA), and a Vision Transformer (ViT) model, for predicting whether ≥C or ≥M-class solar flares occurring within 24 hours. We build a real-time forecasting system using these five models, which can achieve classification and probability forecasting. The 10-fold cross-validation sets are generated in chronological order using the full-disk magnetograms provided by the Solar Dynamics Observatory / Helioseismic and Magnetic Imager at 00:00 UT from May 1, 2010, to March 31, 2023. Then after training, validation, and testing our models, we compare the results with the true skill statistic (TSS) and Brier Skill Score (BSS) as assessment metrics. The major results are as follows: (1) There are no statistically significant differences in TSS and BSS performance between models with attention mechanisms and the CNN model. (2) For ≥C-class flare prediction, the Recall of the ViT model reaches 0.833, significantly better than that of the CNN model. For ≥M-class flare prediction, the Recall of the CNN-ECA and ViT models are 0.799 and 0.855, respectively, which are significantly higher than those of the CNN model. (3) We develop a full-disk solar flare prediction system that has been running since May 1, 2023. By December 31, all five models achieve a TSS of 0.984 for predicting ≥C-class flares, with the CNN-SE model demonstrating a BSS of 0.939. For ≥M-class flares, the CNN-SE model achieves a TSS of 0.304, while the BSS values for the CNN and CNN-SE models are 0.019 and 0.018, respectively. Additionally, the prediction performance for ≥M-class flares on the testing set without No-flare class samples, is similar to that of real-time predictions, validating the good generation performance of the model in real-time forecasting.

Convolutional Neural Networks for Searching Superflares from Pixel-level Data of the Transiting Exoplanet Survey Satellite

Scalable, Advanced Machine Learning-based Approaches for Stellar Flare Identification: Application to TESS short-cadence Data and Analysis of a New Flare Catalogue

Scalable, Advanced Machine Learning Based Approaches for Stellar Flare Identification: Application to TESS Short-cadence Data and Analysis of a New Flare Catalog

FCN4Flare: Fully Convolution Neural Networks for Flare Detection

Superflares on solar-type stars from the first year observation of TESS

Short-Period Variables in TESS Full-Frame Image Light Curves Identified via Convolutional Neural Networks

Solar Flare Prediction with the Hybrid Deep Convolutional Neural Network

Predicting Solar Flares Using CNN and LSTM on Two Solar Cycles of Active Region Data

Hybrid deep convolutional neural network with one-versus-one approach for solar flare prediction

A real-time solar flare forecasting system with deep learning methods

Superflares, Chromospheric Activities, And Photometric Variabilities Of Solar-Type Stars From The Second-Year Observation Of Tess And Spectra Of Lamost

NotPlaNET: Removing False Positives from Planet Hunters TESS with Machine Learning

A machine learning approach for computing solar flare locations in X-rays on-board Solar Orbiter/STIX

Application of Deep Learning Methods for Distinguishing Gamma-Ray Bursts from Fermi/GBM TTE Data

Identifying Exoplanet Candidates Using WaveCeptionNet

Explorer Identification of Low Surface Brightness Tidal Features in Galaxies Using Convolutional Neural Networks

Detecting and Classifying Flares in High-resolution Solar Spectra with Supervised Machine Learning

Identifying Transients in the Dark Energy Survey using Convolutional Neural Networks