Estudio de la mejora de modelos de comportamiento de variables energéticas mediante Committee Machine de redes neuronales
- Alain Porto 1
- Mikel Larrea 2
- Eloy Irigoyen 2
-
1
IK4-IDEKO
info
-
2
Universidad del País Vasco/Euskal Herriko Unibertsitatea
info
Universidad del País Vasco/Euskal Herriko Unibertsitatea
Lejona, España
- Inés Tejado Balsera (coord.)
- Emiliano Pérez Hernández (coord.)
- Antonio José Calderón Godoy (coord.)
- Isaías González Pérez (coord.)
- Pilar Merchán García (coord.)
- Jesús Lozano Rogado (coord.)
- Santiago Salamanca Miño (coord.)
- Blas M. Vinagre Jara (coord.)
Editorial: Universidad de Extremadura
ISBN: 978-84-9749-756-5, 978-84-09-04460-3
Año de publicación: 2018
Páginas: 937-944
Congreso: Jornadas de Automática (39. 2018. Badajoz)
Tipo: Aportación congreso
Objetivos de desarrollo sostenible
Resumen
The present work is based on the modeling of nonlinear dynamic systems using different techniques. The neural models of two complex systems will be presented, such as the consumption of natural gas and electricity. As the work has been carried out with the collaboration of EDP Spain, the company has determined a series of guidelines when developing the study, in such a way that they cover different needs for each of the systems. The main objective of this work is to study which work methodology best fits in the modeling process of energy systems, which have the character of nonlinear dynamic systems, through the use of Artificial Neural Networks, as well as the search for new techniques that can be added to predictive tools that improve their performance, such as Comitte Machine and Boosting. The results presented will show the improvements achieved in the estimation of the energetic variables by means of these techniques.