Solar Irradiance Modeling Using Feed-Forward Neural  Network and Multiple Linear Regression Over Bahir Dar

Tigist Sisay

dc.contributor.author	Tigist Sisay
dc.date.accessioned	2023-01-05T07:14:27Z
dc.date.available	2023-01-05T07:14:27Z
dc.date.issued	2022-12
dc.identifier.uri	http://ir.bdu.edu.et/handle/123456789/14864
dc.description.abstract	Solar irradiance model can reduce the uncertainty of solar power plant output caused by solar irradiance intermittency. In this study, multiple linear regression (MLR) and artificial neural network (ANN) methods were used to predict the direct normal irradiance (DNI) over Bahir Dar. The ANN was developed using the Bayesian learning algorithm. To do this we have utilized hourly meteorological satellite data from National Solar Radiation Data Base (NSRDB) from January 1, 2017 to December 31, 2019. The model was trained using 70% of the data and 30% for testing the accuracy of its prediction. Hour, temperature, relative humidity, solar zenith angle, wind direction, wind speed, diffuse horizontal irradiance and global horizontal irradiance are used as the input variables for models to get DNI. Both MLR and ANN models’ prediction accuracy was assessed statistically using average root-mean-squared error (RMSE), mean absolute error (MAE) and the square of Pearson correlation coefficient (R2 ). The MLR model showed poor generalization abilities with the squared correlation coefficient of R2 = 0.9254 compared to ANN with squared correlation coefficient of R 2~1. In general, the average RMSE and MAE between modeled and observed DNI for ANN model is generally lower compared to multiple regression model. The statistical analysis suggests the potential of ANN over MLR model for DNI predicting using the meteorological data as the main driver of its variability.	en_US
dc.language.iso	en_US	en_US
dc.subject	Physics	en_US
dc.title	Solar Irradiance Modeling Using Feed-Forward Neural Network and Multiple Linear Regression Over Bahir Dar	en_US
dc.type	Thesis	en_US