Long-Term Global Solar Radiation Prediction in 25 Cities in Morocco Using the FFNN-BP Method

Abstract

This article presents different combinations of input parameters based on an intelligent technique, using neural networks to predict daily global solar radiation (GSR) for twenty-five Moroccan cities. The collected measured data are available for 365 days and 25 stations around Morocco. Different input parameters are used, such as clearness index KT, day number, the length of the day, minimal temperature Tmin, maximal temperature Tmax, average temperature Taverage, difference temperature ΔT, ratio temperature T-Ratio, average relative humidity RH, solar radiation at the top outside atmosphere TOA, average wind speed Ws, altitude, longitude, latitude, and solar declination. A different combination was employed to predict daily GSR for the considered locations in order to find the most adequate input parameter that can be used in the prediction procedure. Several statistical metrics are applied to evaluate the performance of the obtained results, such as coefficients of determination (R2), mean absolute percentage error (MAPE), root mean square error (RMSE), normalized root mean square error (NRMSE), mean bias error (MBE), test statistic (TS), linear regression coefficients (the slope “a” and the constant “b”), and standard deviation (σ). It is found that the usage of input parameters gives highly accurate results in the artificial neural network (FFNN-BP) model, obtaining the lowest value of the statistical metrics. The results showed the best input of 25 locations, 12 inputs for Er-Rachidia, Marrakech, Medilt, Taza, Oujda, Nador, Tetouan, Tanger, Al-Auin, Dakhla, Settat, and Safi, seven inputs for Fes, Ifrane, Beni-Mellal, and Meknes, six inputs for Agadir and Rabat, five inputs for Sidi Ifni, Essaouira, Casablanca and Kenitra, four inputs for Ouarzazate, Lareche, and Al-Hoceima. In terms of accuracy, R2 of the selected best inputs parameters varies between 0.9860% and 0.9920%, the range value of MBE (%) being from −0.1076% to −0.5931%, the RMSE between 0.1990 and 0.4580%, the range value of the NRMSE between 0.0355 and 0.8938, and the lowest value MAPE between 0.0019 and 0.0060%. This technique could be used to predict other parameters for locations where measurement instrumentation is unavailable or costly to obtain.

A Comprehensive Methodology for the Statistical Characterization of Solar Irradiation: Application to the Case of Morocco

“…With a larger forecasting term, the daily GHI is predicted using ANN models for 25 Moroccan cities in [31], with empirical and machine learning models for 5 Moroccan cities in [32] and with hybrid ARIMA-ANN model for 3 cities in Morocco in [33]. The daily GHI is also forecasted with ANN models for 35 Moroccan, Algerian, Spanish and Mauritian cities in [34] and the monthly mean daily GHI using time series models in [35].…”

“…[9] Intra-hour GHI cloud retrieval technique to develop a physics-based smart persistence model [10] Intra-day GHI and DNI algorithm using cloud physical properties [11] A 15 min GHI forecasting model [12] Hourly-averaged GHI forecasts [13] Hourly GHI and DNI clear-sky irradiance vs. RRTMG physical radiative transfer model [14] Hourly and daily GHI from mesoscale atmospheric weather research forecasting model [15] Hourly GHI with a three-dimensional meteorology-chemistry model including a treatment of aerosols [16] Hourly GHI exponential smoothing model with decomposition methods [17] A 1 min DNI under a probabilistic approach [18] Short-term GHI with hybrid convolutional ANN model with spatiotemporal correlations [19] Short-term GHI and DNI forecasts of a global numerical weather model [20,21] A 5-30 min GHI and DNI with machine learning techniques [22] A 15 min GHI and DNI with machine learning techniques [23] Intra-day GHI with machine learning techniques [24] A 30 min GHI with ANN algorithm [25,26] Hourly GHI ANN models [27] Mean daily GHI with ANN models [28] A 500 ms-5 min GHI based on k-means algorithm [29] A 5-30 min GHI and DNI based on the k-nearest neighbours algorithm [30] A 30 min-5 h GHI Gaussian process regression method [31] Daily GHI with ANN models for for 25 Moroccan cities [32] Daily GHI with empirical and machine learning models for 5 Moroccan cities [33] Monthly mean daily GHI using time series models [34] Daily GHI with hybrid ARIMA-ANN model for 3 cities in Morocco [35] Daily GHI with ANN models for 35 Moroccan, Algerian, Spanish and Mauritian cities [36] Best Practices Handbook for the Collection and Use of Solar Resource Data, selection of potential sites [37] Steps for solar resource assessment, selection of potential sites [38] Solar resource assessment, selection of potential sites [39] Monthly data, ANN models are used to estimate it in Saudi Arabia …”

“…Proving the above-mentioned trend of employing satellite data for the solar resource validation, Wahab et al [43] compared satellite and ground measured data for northern African countries, including Morocco. Likewise, references [31][32][33][34][35] have also been focused on the irradiation forecasting in several cities of Morocco, employing different methodologies.…”

An Approach to Solar Radiation Prediction Using ARX and ARMAX Models

“…That is, among other things, it can be used to help determine the variables of interest to be used in the work. It also provides a mechanism for selection of the best estimators (Belmahdi et al, 2021).…”

“…In general, several models have been used in the literature to estimate and predict solar radiation. Thus, choosing the best model for each type of application becomes a growing challenge (Belmahdi et al, 2021;Suganthi and Samuel, 2012).…”