Title: Artificial neural network based models for forecasting electricity generation of grid connected solar PV power plant

Authors: Imtiaz Ashraf, A. Chandra

Addresses: Center for Energy Studies, Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110016, [EED, AMU Aligarh], India. ' Center for Energy Studies, Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110016, [EED, AMU Aligarh], India

Abstract: This paper presents an artificial neural network (ANN) approach for forecasting the performance of electric energy generated output from a working 25-kWp grid connected solar PV system and a 100-kWp grid connected PV system installed at Minicoy Island of Union Territory of Lakshadweep Islands. The ANN interpolates among the solar PV generation output and relevant parameters such as solar radiation, module temperature and clearness index. In this study, three ANN models are implemented and validated with reasonable accuracy on real electric energy generation output data. The first model is univariate based on solar radiation and the output values. The second model is a multivariate model based on module temperature along with solar radiation. The third model is also a multivariate model based on module temperature, solar radiation and clearness index. A forecasting performance measure such as percentage root mean square error has been presented for each model. The second model, which gives the most accurate results, has been used in forecasting the generation output for another PV system with similar accuracy.

Keywords: artificial neural networks; electrical energy generated output; forecasting; univariate modelling; solar energy; photovoltaics.

DOI: 10.1504/IJGEI.2004.004704

International Journal of Global Energy Issues, 2004 Vol.21 No.1/2, pp.119 - 130

Published online: 07 Jul 2004 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article