期刊名称：International Journal of Computer Science and Network
印刷版ISSN：2277-5420
出版年度：2020
卷号：9
期号：2
页码：19-27
出版社：IJCSN publisher
摘要：Precise predictions of renewable energy sources play a vital role in bringing them into the electric grid. This researchpresents one of the most powerful machine learning algorithms to forecast the hourly global solar radiance. This study utilizes artificialneural networks (ANN) as the machine learning predictor due to its ability to tackle the nonlinear aspects existing in solar data. The typeof the used ANN in this analysis is a multilayer feed-forward back-propagating neural network, denoted (MLFFBPNN). Nevertheless,choosing the ideal set of input variables, known as features, to train the predictive models created, which are typically user-determined, isa continuing, primary obstacle in obtaining high predictive efficiency. Therefore, this study's precise purpose is to forecast globalhorizontal irradiance by building models of neural networks whose input variables are optimally and systemically chosen by the BorutaAlgorithm, a powerful feature selection method. Prediction models were built based on real-world solar data collected for a site known asBuraydah in Saudi Arabia. For the creation of the developed forecasting models, thirteen features of solar data are considered, includingmonth of the year, day of the month, hour of day, air temperature, relative humidity, surface pressure, wind speed at 3 meters, winddirection, peak wind direction at 3 meters, diffuse horizontal irradiance, direct normal irradiance, azimuth angle, and solar zenith angle.The performance of the suggested models was assessed using four of the most common measures of error. the results stress theimportance of using feature selection techniques when using computational intelligence models to achieve precise solar radiationpredictions..
