Neural architectures for global solar irradiation and air temperature prediction

Authors:
Pierrick Bruneau;Laurence Boudet;Cécilia Damon
Affiliations:
Information, Models & Learning Laboratory, CEA, LIST, Gif sur Yvette CEDEX, France;Information, Models & Learning Laboratory, CEA, LIST, Gif sur Yvette CEDEX, France;Information, Models & Learning Laboratory, CEA, LIST, Gif sur Yvette CEDEX, France
Venue:
ICANN'12 Proceedings of the 22nd international conference on Artificial Neural Networks and Machine Learning - Volume Part II
Year:
2012

Citing 2
Cited 1

NETLAB: algorithms for pattern recognition

NETLAB: algorithms for pattern recognition
Trend Time–Series Modeling and Forecasting With Neural Networks

IEEE Transactions on Neural Networks

Bayesian variable selection in neural networks for short-term meteorological prediction

ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part IV

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a study on neural architectures for the prediction of global solar irradiation and air temperature time series, a useful task for thermal energy management systems. In this contribution, the highly cyclic nature of the variables is carefully considered in the normalization step and the neural architecture design. The standard neural approach is confronted to the absolute daily and the absolute tri-hourly architectures for the prediction of the next 24 hours. For generalization purpose, models are assessed and compared on data from two sites in France. Results show that the absolute models outperform the reference model and some naive models. A complexity analysis also outlines the interest of the proposition.