The automatic detection method of the burst types on the big data of the Chinese Solar Broadband Radio Spectrumeters based on deep learning

jun cheng, National Astronomical Observatories, Chinese Academy of Sciences

This project plans to introduce the deep-learning theory and the related technology to investigate the automatic detection method of the burst types on the big data of the Chinese Solar Broadband Radio Spectrumeters (SBRS). The approximation of complex functions are realized by layer-by-layer nonlinear transformation of the advanced deep network structures, and the pattern features of various burst types and fine structures are automatically learned. Then they automatically marks the type of explosion, and the corresponding burst parameters, like radiation intensity, frequency, bandwidth, life, shape, frequency drift rate, source location and other. This research will further improve the accuracy of solar radio burst parameter extraction, and the understanding of solar radio burst theory.