Clustering via Self-Organizing Maps for solar flare forecasting

• Grant number: 17/07159-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: June 01, 2017
• End date: December 31, 2017
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Guilherme Palermo Coelho
• Grantee: Bruno Antunes Carneiro da Silva
• Host Institution: Faculdade de Tecnologia (FT). Universidade Estadual de Campinas (UNICAMP). Limeira , SP, Brazil

Abstract

Solar flares have become a subject of great interest to researchers from different areas, since such phenomenon has direct impact on terrestrial life. Among these impacts it is possible to mention the interference caused in satellites and radars, the overheating of power transformers of energy distribution networks and even the compromise of the cardiovascular system of terrestrial animals, among others. Given such context, it is important that methods capable of predicting this phenomenon are developed so that preventive measures can be taken to mitigate its effects. Thus, the present project seeks, through the clustering of solar flare data via self-organizing maps, to identify common aspects of the data in the different clusters that will be obtained. In addition, solar flare predictors will be trained with data from each previously identified clusters, so that they can be later used to provide a unique prediction. The goal of such approach is to verify whether predictors representing specific parts of the problem can lead to improvements of the global prediction of solar flares. In other words, this work intends to evaluate whether previous data clustering helps or not the prediction of solar flares. The results obtained here will be compared with those reported in the literature and in other works under development by the High Performance Intelligent Decision Systems (HighPIDS - FT / Unicamp) group, where this work will be developed.