Machine Learning Algorithms Applied to Telemetry Data of SCD-2 Brazilian Satellite

With the advancement of information technologies, big data and data storage and processing capacity, there is an increase in studies on machine learning in different contexts, especially in the spatial context. Specifically talking about artificial satellites, machine learning algorithms can be applied for different purposes, for example to identify the satellite's operating conditions and to predict undesirable situations. The work calculated the performance of six supervised machine learning algorithms in the analysis of telemetry data from the Brazilian satellite SCD2. Five experiments were performed for each supervised machine learning algorithm. To evaluate the algorithms, the following metrics were used: mean squared error (RMSE), coefficient of determination (R2) and mean absolute error (MAE). The Support Vector Machine (SVM) and Bagging Regressor (BR) algorithms obtained better results in the evaluation.