Search results

This study aims to estimate the electric power production of the 20 MWp solar photovoltaic (PV) plant installed in the Adrar region, South of Algeria using minimal knowledge about weather conditions.

In this study, simulation models based on linear and nonlinear approaches were used to estimate accurate energy production from minimum radiometric and meteorological data. Simulations have been carried out by using multiple linear regression (MLR) and artificial neural network (ANN) models with three basic types of neuron connection architectures, namely, feed-forward neural network, cascade-forward neural network (CNN) and Elman neural network. The performance is measured based on evaluation indexes, namely, mean absolute percentage error, normalized mean absolute error and normalized root mean square error.

A comparison of the proposed ANN models has been made with MLR models. The performance analysis indicates that all the ANN-based models are superior in prediction accuracy and stability, and among these models, the most accurate results are obtained with the use of CNN-based models.

The considered model will be adopted in solar PV forecasting areas as part of the operational forecasting chain based on numerical weather prediction. It can be an effective and powerful forecasting approach for solar power generation for large-scale PV plants.

The operational forecasting system can be used to generate an effective schedule for national grid electricity system operators to ensure the sustainability as well as favourable trading performance in the electricity, such as adjusting the scheduling plan, ensuring power quality, reducing depletion of fossil fuel resources and consequently decreasing the environmental pollution.

The proposed method uses the instantaneous radiometric and meteorological data in 15-min time interval recorded over the two years of operation, which made the result exploits a fact that the energy production estimation of PV power generation station is comparatively more accurate.

This paper aims to propose an approach based on physical model integration for surface and cloud albedo computation using an approximate form of the atmospheric radiative transfer equation and sun-pixel-satellite.

The data used in this study are global irradiance collected from for various sites in Algeria, and data were obtained from the processing of the high-resolution visible images taken by the Meteosat Second Generation satellite in 2010.

The results suggest that the standard deviation obtained with this method is similar to that obtained with current estimation methods. The hourly and daily correlation coefficients range between 0.95 and 0.97 and between 0.97 and 0.99, respectively. The hourly and daily mean bias errors range between −0.2 and +1.2 per cent and between −0.2 and +1.4 per cent, respectively. The hourly and daily root mean square errors range between 10 and 17 per cent and between 4 and 8 per cent, respectively.

This paper developed a new estimating method that derives the hourly global horizontal solar irradiation at a ground level from geostationary satellite data under local climate conditions.