This paper presents a comprehensive study on the prediction of influent flow at the Tucuruí Hydroelectric Power Plant, located in the Tocantins-Araguaia basin. The research encompasses five distinct scenarios, varying the architecture of prediction models by incorporating Long Short-Term Memory Recurrent Neural Networks (LSTM) and Convolutional Neural Networks (CNN). The implementation, conducted in Python with the assistance of libraries such as Pandas and NumPy, utilizes historical data of influent flows provided by the National Electric System Operator (ONS) from the Tucuruí, Estreito, and Lajeado power plants. The results obtained were meticulously evaluated
through in-depth analyses, regression metrics, and graphical representations, unequivocally
demonstrating the effectiveness of these approaches in predicting the daily influent flow at the UHE-Tucuruí over temporal horizons ranging from 1 to 7 days. In addition to its methodological contributions, this study provides crucial insights that have the potential to enhance the accuracy of hydrological forecasting, a field of utmost importance in the management of water resources and energy.
|
