@MastersThesis{Muralikrishna:2009:PrÍnGe,
               author = "Muralikrishna, Amita",
                title = "Previs{\~a}o do {\'{\i}}ndice geomagn{\'e}tico dst utilizando 
                         redes neurais artificiais e {\'a}rvore de decis{\~a}o",
               school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
                 year = "2009",
              address = "S{\~a}o Jos{\'e} dos Campos",
                month = "2009-02-13",
             keywords = "Redes neurais artificiais, clima espacial, DST, tempestade 
                         magn{\'e}tica, {\'a}rvore de decis{\~a}o, perceptron 
                         m{\'u}ltiplas camadas, backpropagation, artificial neural 
                         networks, space weather, magnetic storm, decision tree, multilayer 
                         perceptron.",
             abstract = "A Terra sofre constante influ{\^e}ncia da atividade solar 
                         atrav{\'e}s do vento solar, que traz consigo estruturas 
                         resultantes, principalmente, de eventos solares como 
                         explos{\~o}es solares e eje{\c{c}}{\~o}es coronais de massa. A 
                         intera{\c{c}}{\~a}o quase est{\'a}tica do vento solar com o 
                         campo geomagn{\'e}tico forma a estrutura denominada magnetosfera, 
                         que funciona como um escudo, que protege o planeta do plasma 
                         provindo do Sol. No entanto, em fun{\c{c}}{\~a}o das 
                         caracter{\'{\i}}sticas que as estruturas de origem solar 
                         adquirem ao longo do meio interplanet{\'a}rio, pode haver 
                         penetra{\c{c}}{\~a}o de parte dessa mat{\'e}ria para dentro da 
                         magnetosfera. Como conseq{\"u}{\^e}ncia, diversos tipos de 
                         dist{\'u}rbios podem ser gerados no planeta, como, por exemplo, 
                         as auroras e as tempestades geomagn{\'e}ticas, as quais podem 
                         ocasionar diversos danos aos sistemas tecnol{\'o}gicos, entre 
                         outros preju{\'{\i}}zos. Este trabalho aborda a 
                         rela{\c{c}}{\~a}o entre as caracter{\'{\i}}sticas do meio 
                         interplanet{\'a}rio durante o avan{\c{c}}o de estruturas 
                         interplanet{\'a}rias em dire{\c{c}}{\~a}o {\`a} Terra e os 
                         efeitos sentidos pelo campo geomagn{\'e}tico, como resposta a 
                         essas caracter{\'{\i}}sticas. O foco principal {\'e} a 
                         previs{\~a}o do comportamento do campo geomagn{\'e}tico, medido, 
                         neste trabalho, pelo {\'{\i}}ndice geomagn{\'e}tico Dst, 
                         levando-se em conta, principalmente, as tr{\^e}s coordenadas do 
                         campo magn{\'e}tico interplanet{\'a}rio. As ferramentas 
                         escolhidas para resolver o problema n{\~a}o-linear foram as 
                         t{\'e}cnicas: Rede Neural Artificial do tipo Perceptron de 
                         M{\'u}ltiplas Camadas, treinada com algoritmo backpropagation, 
                         Mapa Auto-organiz{\'a}vel de Kohonen e {\'A}rvore de 
                         Decis{\~a}o com algoritmo J48. Foi poss{\'{\i}}vel comprovar 
                         algumas rela{\c{c}}{\~o}es e questionar a exist{\^e}ncia de 
                         outras com a {\'A}rvore de Decis{\~a}o e prever, com {\'o}timo 
                         percentual de efici{\^e}ncia, o {\'{\i}}ndice geomagn{\'e}tico 
                         Dst com a Rede MLP. ABSTRACT: The Earth suffers constant influence 
                         of the solar activity through the solar wind, which brings with it 
                         the resulting structures, mainly phenomena like solar flares and 
                         coronal mass ejection. The almost static interaction between the 
                         solar wind and the geomagnetic field forms a structure called 
                         magnetosphere, which acts as a shield that protects the planet 
                         from radiation and solar plasma. However, depending on the 
                         characteristics that these structures of solar origin acquire 
                         throughout the interplanetary medium, a part of the energy and 
                         matter may penetrate into the magnetosphere. As a result, 
                         different types of disturbances can be generated on the planet, 
                         for example, the aurora and the geomagnetic storms, which can 
                         cause damage to various technological systems, among other losses. 
                         The present work formulates the relationship between the 
                         characteristics of the interplanetary medium during the progress 
                         of the interplanetary structures towards the Earth and the effects 
                         observed on the geomagnetic field, in response to these 
                         characteristics. The main focus is on forecasting the behavior of 
                         the geomagnetic field, represented in this work by the Dst index, 
                         using for that, mainly, the three interplanetary magnetic field 
                         components. The tools chosen here to solve the non-linear problem 
                         were the Multi-layer Perceptrons Artificial Neural Network, 
                         trained with the backpropagation algorithm; the Kohonen 
                         Self-Organizing Map and the Decision Tree with the J48 algorithm. 
                         It was possible to establish some relationships and to question 
                         the existence of others with the Decision Tree, and predict the 
                         geomagnetic Dst index with great percentage efficiency with the 
                         Artificial Neural Network.",
            committee = "Rosa, Reinaldo Roberto (presidente) and Silva, Jos{\'e} 
                         Dem{\'{\i}}sio Sim{\~o}es da (orientador) and Lago, Alisson Dal 
                         (orientador) and Alarcon, Walter Demetrio Gonzalez and 
                         Os{\'o}rio, Fernando Santos",
           copyholder = "SID/SCD",
         englishtitle = "Geomagnetic DST index forecast using artificial neural networks 
                         and decision tree",
             language = "pt",
                pages = "132",
                  ibi = "8JMKD3MGP8W/355JR6S",
                  url = "http://urlib.net/rep/8JMKD3MGP8W/355JR6S",
           targetfile = "publicacao.pdf",
        urlaccessdate = "26 fev. 2021"
}