Aplicação de aprendizagem de máquina para previsão de energia eólica gerada

Abstract

Artificial intelligence (AI) has experienced significant advances in recent years, boosting its growing field of application in several areas, among which the application in predictive systems stands out. In these systems, previously collected data are mapped and labeled, and these data are used as input for AI algorithms, which aim to predict possible future values. For the present work, an area was chosen, which has been standing out in Brazil and in the world, which deals with the generation of energy through wind generators. The present work explores the concepts and functioning of predictive models and proposes to carry out an exploratory research, in order to identify the main methods that are used, through a bibliographical research and a systematic review. From this research, it is proposed the implementation of two predictive models most used in the area of regression problems, namely the Long Short Term Memory (LSTM) and the Extreme Gradient Boosting (XGBoost) models. After the implementation and training of the proposed models, the results quantified by metrics, known and widely used to evaluate the values obtained in the prediction of the energy generated by these models, are presented. [resumo fornecido pelo autor]