Analysis of the multi-layer Perceptrons applied to solar flares forecasting

• Grant number: 15/25456-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: March 01, 2016
• End date: December 31, 2016
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: André Leon Sampaio Gradvohl
• Grantee: Thaís Veinert Teche
• Host Institution: Faculdade de Tecnologia (FT). Universidade Estadual de Campinas (UNICAMP). Limeira , SP, Brazil

Abstract

Space weather effects, i.e. the occurrence of solar phenomena with effects on Earth can affect various technologies. For instance, extreme solar flares, class X, can affect satellite orbits, cause disruptions on the power distribution in power grids, radio waves disturbances and navigation systems problems, among others. Therefore, the sooner systems preview the occurrence of solar flares, the more likely to mitigate its effects. Thus, this project aims to analyze the Multi-Layer Perceptron strategy used to forecast solar flares, looking for a set of attributes that leads to the highest possible hit rate in these forecasts. In addition, the project proposes the analysis and improvement of a solar flare predictive system, which is currently under development at the High Performance Intelligent Decision Systems (HighPIDS) research group, which is based on the School of Technology at the University of Campinas.