2026-10-15T04:04:06Z https://www.scidb.cn/oai

10.57760/sciencedb.space.03281 2025-11-07T16:23:37Z

2025-11-07 Dataset for the paper "Identification of CME-Driven Shocks Based on Numerical Simulation and Deep Learning" doi:10.57760/sciencedb.space.03281 en Sec2&3_Data: Data for Sections 2 and 3 of the paper, used for training and testing the CNN model. The folder includes:(1) 3D CME numerical simulation data, selected using the MHD method with the GCS-based flux-rope model as the CME initialization model. It simulates the propagation and evolution of CME in a non-uniform background solar wind.(2) CME-driven shock prediction data based on CNN methods. The file predict.xlsx contains the evaluation results of the CNN model, including BCE loss for each CME simulation time, and a comparison of execution times between CNN and traditional methods.(3) training_log.csv is the model training log, including the loss rates for the training and testing datasets for each epoch, the training time for a single epoch, total training time, and learning rate updates.Sec4_DATA: Data for Section 4 of the paper, used for validating the CNN method. The folder includes:(1) 3D MHD numerical simulation data of the ICME event on December 4, 2021. The CME initial model is a linear force-free spheromak model. The data is divided into files for the corona region and the interplanetary region.(2) CME-driven shock prediction data based on CNN methods. This includes data files for the corona region (01h, 03h, 05h) and the interplanetary region (20h, 30h, 40h, 50h, 60h, 70h, 80h). Solar coronal mass ejection shocks; Magnetohydrodynamical simulations; Convolutional neural networks JING'EN LI FANG SHEN YI YANG MENGXUAN MA LIPING YANG RESTRICTED dataset Science Data Bank