Enhancing Solar Forecasting Accuracy: A Comprehensive Evaluation of Artificial Neural Networks for Global Horizontal Irradiance Prediction
Abstract views: 7
/ 
PDF downloads: 7
DOI:
https://doi.org/10.5281/zenodo.15038442Keywords:
Solar Energy, Artificial Neural Networks, Global Horizontal Irradiance Forecasting, MLP, NARXAbstract
Solar energy plays a pivotal role in the global transition toward sustainable energy systems,
providing a clean and renewable power source. However, the inherent variability of solar irradiance
presents significant challenges for energy management and grid stability. Accurate forecasting of Global
Horizontal Irradiance (GHI) is crucial for optimizing photovoltaic (PV) power generation and ensuring a
reliable energy supply. GHI prediction is particularly complex due to its dependence on dynamic
meteorological factors, including cloud cover, atmospheric aerosols, temperature, and humidity.
Traditional statistical and physical models often struggle to capture these nonlinear patterns, whereas
artificial intelligence (AI)-based approaches, particularly artificial neural networks (ANNs), have
demonstrated significant potential in improving forecasting accuracy. This study examines GHI
prediction in Dakhla City, Morocco, utilizing two AI-based models: the Multilayer Perceptron (MLP) and
the Nonlinear Autoregressive Model with Exogenous Inputs (NARX). The objective is to enhance
forecasting accuracy to facilitate more efficient solar energy integration. A performance evaluation based
on statistical metrics reveals that the NARX model significantly outperforms the MLP model, achieving a
regression coefficient (R) of 0.999 and a root mean square error (RMSE) of 8.722. This superior
performance is attributed to the NARX model’s capacity to capture nonlinear dependencies and
incorporate past values alongside exogenous inputs. These findings underscore the effectiveness of AI
driven models in solar energy forecasting. Enhanced GHI predictions can contribute to improved grid
stability, optimized solar energy utilization, and the advancement of Morocco’s renewable energy
objectives. As such, AI-based forecasting emerges as a critical tool for sustainable energy management.
Downloads
References
V. Gayathry, D. Kaliyaperumal, et S. R. Salkuti, « Seasonal solar irradiance forecasting using artificial intelligence techniques with uncertainty analysis », Sci Rep, vol. 14, no 1, p. 17945, août 2024, doi: 10.1038/s41598-024-68531-3.
H. N. Nguyen, Q. T. Tran, C. T. Ngo, D. D. Nguyen, et V. Q. Tran, « Solar energy prediction through machine learning models: A comparative analysis of regressor algorithms », PLoS ONE, vol. 20, no 1, p. e0315955, janv. 2025, doi: 10.1371/journal.pone.0315955.
S. Serag et A. Echchelh, « Technical and Economic Evaluation of Electricity Generation and Storage Using Renewable Energy Sources on Socotra Island, Yemen », Iraqi Journal of Science, p. 2809‑2842, juin 2023, doi: 10.24996/ijs.2023.64.6.14.
S. A. Syed, W. B. Chang, H. Nisar, H. N. Riaz, K. H. Yeap, et N. M. Zaber, « Short-term solar irradiance forecasting using deep learning models », E3S Web Conf., vol. 603, p. 03003, 2025, doi: 10.1051/e3sconf/202560303003.
K. Barhmi, C. Heynen, S. Golroodbari, et W. Van Sark, « A Review of Solar Forecasting Techniques and the Role of Artificial Intelligence », Solar, vol. 4, no 1, p. 99‑135, févr. 2024, doi: 10.3390/solar4010005.
P. Kumari et D. Toshniwal, « Deep learning models for solar irradiance forecasting: A comprehensive review », Journal of Cleaner Production, vol. 318, p. 128566, oct. 2021, doi: 10.1016/j.jclepro.2021.128566.
S. Serag, A. Echchelh, et B. Morrone, « Hydroelectric and Hydrogen Storage Systems for Electric Energy Produced from Renewable Energy Sources », ENERGY, vol. 0, no 0, p. 1‑10, 2024, doi: 10.32604/ee.2024.054424.
Y. Yang, Z. Tang, Z. Li, J. He, X. Shi, et Y. Zhu, « Dual-Path Information Fusion and Twin Attention-Driven Global Modeling for Solar Irradiance Prediction », Sensors, vol. 23, no 17, p. 7469, août 2023, doi: 10.3390/s23177469.
F. ALtalqi, S. Fennane, H. Mabchour, H. Kacimi, et A. Echchelh, « Design monopole antenna of ultra-wideband high bandwidth and high efficiency for ground penetrating radar application », TELKOMNIKA, vol. 22, no 4, p. 812, août 2024, doi: 10.12928/telkomnika.v22i4.25340.
Ü. Ağbulut, A. E. Gürel, et Y. Biçen, « Prediction of daily global solar radiation using different machine learning algorithms: Evaluation and comparison », Renewable and Sustainable Energy Reviews, vol. 135, p. 110114, janv. 2021, doi: 10.1016/j.rser.2020.110114.
R. Elmousaid, N. Drioui, R. Elgouri, H. Agueny, et Y. Adnani, « Ultra-short-term global horizontal irradiance forecasting based on a novel and hybrid GRU-TCN model », Results in Engineering, vol. 23, p. 102817, sept. 2024, doi: 10.1016/j.rineng.2024.102817.
Z. Wang, T. Peng, X. Zhang, J. Chen, S. Qian, et C. Zhang, « Enhancing multi-step short-term solar radiation forecasting based on optimized generalized regularized extreme learning machine and multi-scale Gaussian data augmentation technique », Applied Energy, vol. 377, p. 124708, janv. 2025, doi: 10.1016/j.apenergy.2024.124708.
Z. Wang, Y. Tang, et Z. Zhang, « Dual-branch deep learning architecture for enhanced hourly global horizontal irradiance forecasting », Expert Systems with Applications, vol. 252, p. 124115, oct. 2024, doi: 10.1016/j.eswa.2024.124115.
Oliver O. Apeh et Nnamdi I. Nwulu, « Machine Learning Approach for Short- and Long-Term Global Solar Irradiance Prediction », j. of environ. & earth. sci., vol. 7, no 1, p. 321‑342, déc. 2024, doi: 10.30564/jees.v7i1.7060.
Z. Boussaada, A. Remaci, O. Curea, H. Camblong, et N. Mrabet Bellaaj, « Prediction of the Daily Direct Solar Radiation Using Nonlinear Autoregressive Exogenous (Narx) Network Model », in Renewable Energy Sources, University of Maribor Press, juill. 2017, p. 343‑354. doi: 10.18690/978-961-286-061-5.30.
« Prediction of Hourly Solar Radiation in Amman-Jordan by Using Artificial Neural Networks », IJTEE, vol. 14, no 02, 2018, doi: 10.5383/ijtee.14.02.003.
Y. A. Al-Sbou et K. M. Alawasa, « Nonlinear Autoregressive Recurrent Neural Network Model For Solar Radiation Prediction », vol. 12, no 14, 2017.
L. B. Mohammed, M. A. Hamdan, E. A. Abdelhafez, et W. Shaheen, « Hourly Solar Radiation Prediction Based on Nonlinear Autoregressive Exogenous (Narx) Neural Network », vol. 7, no 1, 2013.
Y. Amellas, O. E. Bakkali, A. Djebli, et A. Echchelh, « Short-term wind speed prediction based on MLP and NARX network models », IJEECS, vol. 18, no 1, p. 150, avr. 2020, doi: 10.11591/ijeecs.v18.i1.pp150-157.
Y. Amellas, Α. Djebli, et Α. Echchelh, « Levenberg-Marquardt Training Function using on MLP, RNN and Elman Neural Network to Optimize Hourly Forecasting in Tetouan City (Northern Morocco) », JESTR, vol. 13, no 1, p. 67‑71, févr. 2020, doi: 10.25103/jestr.131.09.
S. Mohamed, V. Elvally, A. K. Mahmoud, et A. Zoubir, « The prediction of the usual solar irradiation in the Sahel using the artificial neural networks (case study: 50 MW power plant in Nouakchott) », IJPEDS, vol. 15, no 3, p. 1739, sept. 2024, doi: 10.11591/ijpeds.v15.i3.pp1739-1748.
M. W. Gardner et S. R. Dorling, « Artificial neural networks (the multilayer perceptron)—a review of applications in the atmospheric sciences », Atmospheric Environment, vol. 32, no 14‑15, p. 2627‑2636, août 1998, doi: 10.1016/S1352-2310(97)00447-0.
A. Cheng et Y. M. Low, « Improved generalization of NARX neural networks for enhanced metamodeling of nonlinear dynamic systems under stochastic excitations », Mechanical Systems and Signal Processing, vol. 200, p. 110543, oct. 2023, doi: 10.1016/j.ymssp.2023.110543.
Y. Zheng, W. Zhang, J. Xie, et Q. Liu, « A Water Consumption Forecasting Model by Using a Nonlinear Autoregressive Network with Exogenous Inputs Based on Rough Attributes », Water, vol. 14, no 3, p. 329, janv. 2022, doi: 10.3390/w14030329.
C. Ibtissem et L. Nouredine, « A hybrid method based on conjugate gradient trained neural network and differential evolution for non linear systems identification », in 2013 International Conference on Electrical Engineering and Software Applications, Hammamet, Tunisia: IEEE, mars 2013, p. 1‑5. doi: 10.1109/ICEESA.2013.6578397.
R. Elmousaid, N. Drioui, R. Elgouri, H. Agueny, et Y. Adnani, « Accurate short-term GHI forecasting using a novel temporal convolutional network model », e-Prime - Advances in Electrical Engineering, Electronics and Energy, vol. 9, p. 100667, sept. 2024, doi: 10.1016/j.prime.2024.100667.
