RU-net: A Residual U-net for Automatic Interplanetary Coronal Mass Ejection Detection

Abstract Detection methods for interplanetary coronal mass ejections (ICMEs) from in situ spacecraft measurements are mostly manual, which labor-intensive and time-consuming, being prone to the inconsistencies of identification criteria incompleteness existing catalogs. Therefore, automatic detection ICMEs has aroused interest astrophysical community. Of these methods, convolutional neural network–based show advantages fast speed high precision. To further improve computing feasibility performance, this paper proposes a method called residual U-net (RU-net), perspective time-series segmentation. With help architecture, we design an encoder–decoder network with skip connection capture multiscale information, where end-to-end architecture embedded element is formulated accelerate algorithmic convergence. For data 1997 October 1 2016 January collected by Wind spacecraft, results our experiments demonstrate competitive performance proposed RU-net terms accuracy efficiency (178 230 detected test set, F score 80.18%).