Solar flares are explosions on the surface of the Sun caused by sudden changes in the magnetic field of its surface. An explosion happens when the magnetic fields of different polarities around the Sun connect to others and create magnetic arches. Solar flares are classified according to their power. Their consequences can affect orbiting satellites, cause interruptions in the power distribution in the power grid, disturbances to radio waves and navigation systems, among others. Thus, it is important to forecast the occurrence and intensity of solar flares. The x-ray flux is usually part of the set of input parameters of machine learning algorithms of solar flare classification, as well as classification of occurrence and non-occurrence of a solar flare and prediction of solar flares. This project continues a research project of the HighPIDS research group, which used values of X-ray emissions on a MLP artificial neural network to predict peaks of X-ray fluxes. The results were promising with MLP resulting in low prediction error for prediction of X-ray flux in the next minute. This study's main objectives are: 1) improving MLP prediction horizon according to new experiments to be performed; 2) create an interface to trigger warnings of x-ray peaks, indicating possibility of solar flares. The work proposed is innovative the analysis of X-ray fluxes as a reliable parameter for predicting solar flares in isolation, which would allow the incorporation of new parameters gradually to the prediction.
News published in Agência FAPESP Newsletter about the scholarship: