Prediction of Solar Irradiance by Artificial Neuron Networks

EasyChair Preprint 9051

Abstract

Accurate forecasting of solar irradiance is essential to minimize operating costs for solar photovoltaic (PV) generation, as it is often used to predict power output. This work aims at the prediction of daily solar irradiation for a photovoltaic power plant using artificial neural networks. By adopting the multilayer perceptron (MLP) in the Matlab environment, three neuronal structures are studied and compared to the naive model (known as the persistence model). In order to be able to evaluate the performance of the proposed forecasting system, two years of meteorological data (2019-2020) are collected from the Oued Kebrit photovoltaic plant in S/Ahras, and the RMSE and the MAE are used as error indices. For a one day ahead solar irradiance forecasting, the neural network of one hidden layer with 6 neurons and a 5-day prediction lag has shown the best performance

Keyphrases: Artificial Neural Networks, Machine Learning., prediction, solar irradiation

@booklet{EasyChair:9051,
  author    = {Hadia Belhouchet and Khaled Khelil and Zohir Boumous and Samira Boumous},
  title     = {Prediction of Solar Irradiance by Artificial Neuron Networks},
  howpublished = {EasyChair Preprint 9051},
  year      = {EasyChair, 2022}}