Day Ahead Hourly Global Horizontal Irradiance Forecasting—Application to South African Data

Abstract

Due to its variability, solar power generation poses challenges to grid energy management. In order to ensure an economic operation of a national grid, including its stability, it is important to have accurate forecasts of solar power. The current paper discusses probabilistic forecasting of twenty-four hours ahead of global horizontal irradiance (GHI) using data from the Tellerie radiometric station in South Africa for the period August 2009 to April 2010. Variables are selected using a least absolute shrinkage and selection operator (Lasso) via hierarchical interactions and the parameters of the developed models are estimated using the Barrodale and Roberts’s algorithm. Two forecast combination methods are used in this study. The first is a convex forecast combination algorithm where the average loss suffered by the models is based on the pinball loss function. A second forecast combination method, which is quantile regression averaging (QRA), is also used. The best set of forecasts is selected based on the prediction interval coverage probability (PICP), prediction interval normalised average width (PINAW) and prediction interval normalised average deviation (PINAD). The results demonstrate that QRA gives more robust prediction intervals than the other models. A comparative analysis is done with two machine learning methods—stochastic gradient boosting and support vector regression—which are used as benchmark models. Empirical results show that the QRA model yields the most accurate forecasts compared to the machine learning methods based on the probabilistic error measures. Results on combining prediction interval limits show that the PMis the best prediction limits combination method as it gives a hit rate of 0.955 which is very close to the target of 0.95. This modelling approach is expected to help in optimising the integration of solar power in the national grid.

Regularized quantile regression averaging for probabilistic electricity price forecasting

“…Following the unprecedented success in the Global Energy Forecasting Competition 2014, QRA became a popular technique for probabilistic energy forecasting (see, e.g., Zhang et al, 2016;Liu et al, 2017;Zhang et al, 2018;Kostrzewski and Kostrzewska, 2019;Mpfumali et al, 2019;Serafin et al, 2019;Uniejewski et al, 2019;Wang et al, 2019). However, a recent study of Marcjasz et al (2020) has revealed the method's vulnerability to low quality predictors when the set of regressors is larger than just a few.…”

“…Formally introduced by Nowotarski and Weron (2015), Quantile Regression Averaging (QRA) has sparked interest in the EPF community after its unprecedented success in GEFCom2014, where the top two winning teams in the price track - Gaillard et al (2016) and -used variants of QRA. Since the method is a general forecasting technique not limited to electricity prices, a number of authors have reported its successful application Email addresses: bartosz.uniejewski@pwr.edu.pl (Bartosz Uniejewski), rafal.weron@pwr.edu.pl (Rafał Weron) in areas ranging from load (Liu et al, 2017;Zhang et al, 2018;Wang et al, 2019) to wind power (Zhang et al, 2016) and irradiance forecasting (Mpfumali et al, 2019).…”

Inverter Efficiency Analysis Model Based on Solar Power Estimation Using Solar Radiation

“…Solar radiation is a primary factor affecting power output. Some studies are ongoing with the goal of estimating solar radiation to predict future power output [30][31][32]. There are also studies on power output estimation based on ambient temperature, wind velocity, and incident light [33][34][35].…”

Forecasting the Global Horizontal Irradiance based on Boruta Algorithm and Artificial Neural Networks using a Lower Cost

“…The obtained results showed a very high accuracy. Besides, in South Africa [34], a research study carried out for discussing probabilistic of forecasting the GHI before 24 hours, using two machine learning methods and the data collected during the period from 2009 to 2010. The study gave excellent results but not exceeded by 95.5%.…”