Predição da cintilação ionosférica de amplitude pela aplicação de métodos de aprendizagem de máquina.

• Ano de defesa: 2021
• Autor(a) principal: CARVALHO, Otávio
• Orientador(a): BARRADAS FILHO, Alex Oliveira
• Banca de defesa: BARRADAS FILHO, Alex Oliveira , CARVALHO, André Carlos Ponce de Leon Ferreira de , SOUZA, Francisco das Chagas de , FONSECA NETO, João Viana da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Maranhão
• Programa de Pós-Graduação: PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA AEROESPACIAL/CCET
• Departamento: DEPARTAMENTO DE ENGENHARIA DA ELETRICIDADE/CCET
• País: Brasil
• Palavras-chave em Português: Comunicação e navegação via satélite; Cintilação ionosférica; Modelos preditivos; Aprendizagem de máquina
• Palavras-chave em Inglês: Satellite communication and navigation; Ionospheric scintillation; Predictive models; Machine learning
• Área do conhecimento CNPq: Engenharia Aeroespacial
• Link de acesso: https://tedebc.ufma.br/jspui/handle/tede/3650

Today’s society is increasingly dependent on products and services based on the use of satellites, among which satellite communication and navigation systems stand out, technologies directly affected by the phenomenon of ionospheric scintillation, which can compromise or even make unfeasible the use of such systems. In this context, it becomes important to develop tools capable of predicting the occurrence of ionospheric scintillation, however, the modeling of this phenomenon is quite complex due to the influence of several other aspects. Therefore, the main objective of this work is to develop short-term predictive models, both quantitative and qualitative, about amplitude ionospheric scintillation. For this, machine learning techniques will be used, considering information related to geomagnetic activity, temporal and geographic dependence of the phenomenon, solar and interplanetary activities and the state of the ionosphere, evaluating the influence of different attributes on the performance of the models obtained. The methodology used was based on the use of machine learning techniques, both regression and classification, highlighting the algorithms Random Forest, logistic regression, and multiple linear regression. The main results obtained are related to the prediction of ionospheric scintillation 30 minutes in advance, in addition to aspects related to the analysis of the attributes used. Finally, the future directions of the research carried out are indicated.