Neural network forecasting of energy generation by solar panels
Abstract
The main purpose of this paper is analysis of various regression methods application on quality of short-term solar PV forecasting. Multilayer perceptron and decision trees were chosen in order to solve prediction problem. Real historical data on solar PV forecasting are used as experimental datasets. The best coefficient of determination was 0.94.
About the Authors
S. M. Stepanov
Belarusian state university of informatics and radioelectronics
Belarus
Belarus
N. A. Iskra
Belarusian state university of informatics and radioelectronics
Belarus
Belarus
References
1. Huang R., Huang T., Gadh R. Solar Generation Prediction using the ARMA Model in a Laboratory-level Micro-grid // IEEE Third International Conference on Smart Grid Communications. 2012. P. 528-533.
2. Bacher P. Short-term Solar Power Forecasting // Technical University of Denmark. 2008. P. 109.
3. Leung I.J.H., Jordan J.E. Image processing for weather satellite cloud segmentation // Electrical and Computer Engineering. 1996. P. 953-956.
4. Fine-grained photovoltaic output prediction using Bayesian Ensemble / Prithwish Chakraborty [et al.] // AAAI. 2012. P. 7.
5. Хайкин С. Нейронные сети. Полный курс. М.: Вильямс, 2006. 1104 с.
6. Tom M.M. Machine Learning. Boston, Massachusetts, 1997. 52 p.
7. Patrick K.W., Sonja S.T. A practical introduction to econometric methods: classical and modern. University of the West Indies Press, 2002. 307 p.
8. Anaconda [Электронный ресурс]. - URL: https://www.anaconda.com (дата обращения: 01.12.2017).
9. Diederik P.K., Jimmy B.A. A method of stochastic optimization // 3rd ICLR. San Diego, 2015. P. 15.
10. Friedman J.H. Greedy Function Approximation: A Gradient boosting machine // IMS Reitz Lecture. 1999. P. 44.
Review
For citations:
Stepanov S.M., Iskra N.A. Neural network forecasting of energy generation by solar panels. Doklady BGUIR. 2018;(3):26-31. (In Russ.)
Views: 349
Email this article 