Análisis predictivo para minería de datos y proyección a corto plazo de la demanda de potencia en el sistema eléctrico ecuatoriano

Abstract

Proper planning of the electricity sector makes it possible to guarantee the supply of electricity demand, either in traditional energy conditions or even in those classified as adverse, with the incorporation of electricity projects that use renewable and efficient energy sources for the short, medium, and long term. The former ARCONEL, now the Agency for the Regulation and Control of Energy and Non-Renewable Natural Resources (ARCERNNR), as the entity in charge of regulating the electricity sector, carries out technical-economic support studies for the projection of electricity demand. The need to implement actions to manage demand through regulatory and rate schemes to optimize the operation of the National Interconnected System makes it necessary to analyze the forecast of electrical power demand curves with hourly resolution. The main objective of the study carried out in this Degree Project is to implement a computational tool in Python programming language, using machine learning algorithms: Light Gradient Boosting Machine, Random Forest Regressor, Decision Tree Regressor, and the traditional statistical algorithm ARIMA, in a Proposed procedure for data mining tasks and short-term projection of hourly demand for electrical power.