Closed recurring units (GRU) vs artificial neural networks in the prediction of electricity generation at the Illuchi Hydroelectric Power Plant

Abstract

The prediction of events has been, since ancient times, a phenomenon capable of generating curiosity in human beings. However, to achieve a projection of a future event, a detailed analysis of data is required to predict subsequent events. Objectives: The objective of the research was to develop two prediction systems using artificial neural networks and GRU to determine the predicted electrical generation at the ILLUCHI HYDROELECTRIC PLANT. Methodology: The data used for this study were collected from ELEPCO S.A. operators based on the years 2008 to 2020. The input variables were the date and the energy generated to develop different cases with different conditions to arrive at a successful Recurrent Neural Network model. Results: Once the variables of the model were understood, the data was divided into two groups: training 70% and validation 30% respectively. The ADAM algorithm and libraries provided by Python were used for the corresponding training. General area of study: Engineering. Specific area of study: Electrical.