Analysis and Prediction of Droughts in the Hydroelectric Power Plants of Ecuador Using Machine Learning

In Ecuador, electricity generation relies on 90% hydroelectric sources, which are highly vulnerable to climate variability and phenomena like El Niño. Water scarcity in critical watersheds, such as Paute and Coca, compromises energy stability and forces the use of more expensive thermoelectric plants. This project addresses this issue by implementing machine learning models to predict droughts based on historical precipitation, temperature, and humidity data provided by INAMHI. The methodology includes the collection and preprocessing of climate data from the last decade, followed by the training and evaluation of algorithms such as K-Means and Artificial Neural Networks (ANN). The use of error metrics like MAE, RMSE, and MSE will allow for the validation of the accuracy of the developed models. This technological approach is expected to provide a robust tool for preventive decision-making in reservoir management. The results will contribute to the optimization of national energy production, reducing dependence on fossil fuels and strengthening the resilience of the electrical system against extreme weather events.<br/><br/><b>Goal</b>: <br/>Propose a predictive model using machine learning techniques to anticipate drought events in the watersheds of Ecuador's hydroelectric power plants. The goal is to improve water resource management and mitigate negative effects on electricity production.<br/><br/><b>Research lines</b>: <br/>Big data and analytics