A New Technique for Solar Activity Forecasting Using Recurrent Elman Networks

Authors: Salvatore Marra, Francesco C. Morabito

Abstract:

In this paper we present an efficient approach for the prediction of two sunspot-related time series, namely the Yearly Sunspot Number and the IR5 Index, that are commonly used for monitoring solar activity. The method is based on exploiting partially recurrent Elman networks and it can be divided into three main steps: the first one consists in a “de-rectification" of the time series under study in order to obtain a new time series whose appearance, similar to a sum of sinusoids, can be modelled by our neural networks much better than the original dataset. After that, we normalize the derectified data so that they have zero mean and unity standard deviation and, finally, train an Elman network with only one input, a recurrent hidden layer and one output using a back-propagation algorithm with variable learning rate and momentum. The achieved results have shown the efficiency of this approach that, although very simple, can perform better than most of the existing solar activity forecasting methods.

Keywords: Elman neural networks, sunspot, solar activity, time series prediction.

Digital Object Identifier (DOI): doi.org/10.5281/zenodo.1084139

References:

[1] Van der Linden, R. A. M. & the SIDC team. Online catalogue of the sunspot index. Available: http://sidc.oma.be/html/sunspot.html
[2] S. Bengio, F. Fessant, D. Collobert, "On the Prediction of Solar Activity using Different Neural Networks Models," Annales Geophysicae, vol. 14, pp.20-26, 1995.
[3] F. Fessant, S. Bengio, D. Collobert, "Use of Modular Architectures for Time Series Prediction," Neural Processing Letters, vol. 3, pp. 101-106, 1996.
[4] Lj. R., Cander, X. Lamming, "Neural Networks in Ionospheric Prediction and Short-Term Forecasting," Proceedings of the IEEE Tenth International Conference on Antennas and Propagation, vol. 2, pp. 14- 17, 1997.
[5] E. A. Wan, "Combining Fossils and Sunspots: Committee Predictions," Proceedings of the IEEE International Conference on Neural Networks, vol. 4, pp. 2176-2180, 1997.
[6] M. G. Genet, A. Pétrowsky, "Sunspot Number Prediction by a Conditional Distribution Discrimination Tree," Proceedings of the IEEE International Joint Conference on Neural Networks, vol. 1, pp. 814-819, 2003.
[7] J. L. Elman, "Finding Structure in Time," Cognitive Science, vol. 14, pp. 179-221, 1990.
[8] R. Bracewell, "Spectral Analysis of the Elatina Series," Solar Physics, vol. 116, pp. 179-197, 1988.
[9] D. Nguyen, B. Widrow, "Improving the Learning Speed of Two Layer Neural Networks by Choosing Initial Values of the Adaptive Weights," Proceedings of the International Joint Conference on Neural Networks, vol. 3, pp. 21-26, 1990.
[10] B.W. Wah, M. L. Qian, M.L, "Time-series Predictions using Constrained Formulations for Neural-Network Training and Cross Validation," Proceedings of the International Conference on Intelligent Information Processing, 16th IFIP World Computer Congress, pp. 220- 226, 2000.
[11] R. Boné, M. Crucianu, "An Evaluation of Constructive Algorithms for Recurrent Networks on Multi-Step Ahead Prediction," Proceedings of the International Conference on Neural Information Processing, pp. 547-551, 2002.
[12] A. S. Weigend, B. A. Huberman, D. E. Rumelhart, "Predicting the Future: a Connectionist Approach," International Journal of Neural Systems, vol. 1, no. 3, pp. 193-209, 1990.
[13] A. Aussem, "Dynamical Recurrent Neural Networks Towards Prediction and Modeling of Dynamical Systems," Neurocomputing, vol. 28, pp. 207-232, 1999.
[14] S. Nowland, G. Hinton, "Simplifying Neural Networks by Soft Weight Sharing," Neural Computation, vol. 4, no. 4 pp. 473-493, 1992.
[15] A. B. Geva, "ScaleNet - Multiscale Neural-Network Architecture for Time Series Prediction," IEEE Transactions on Neural Networks, vol. 9, no. 5, pp. 1471-1482, 1998.