Solar radiation modeling using advanced statistical and machine learning techniques / Muhammed Abdullah Hassan Ahmed ; Supervised Adel Khalil Hassan Khalil , Sayed Ahmed Kaseb , Mahmoud Abdelwahab Kassem

Thesis (Ph.D.) - Cairo University - Faculty of Engineering - Department of Mechanical Power Engineering

In this study, the different solar radiation components (i.e. global, diffuse, normal and tilted radiations) are measured at different solar-meteorological stations in high time resolution (one- or 10-minute time steps) and used to develop new models for all solar radiation components using different machine learning and statistical algorithms. The machine learning algorithms include the multi-layer perceptron (MLP), support vector machines (SVM), adaptive neuro-fuzzy inference system (ANFIS), decision trees (DT), and ensemble methods (gradient boosting, bagging and random forest 2RF3). In addition to these stochastic algorithms, time series techniques have also been considered, including the auto-regressive integrated moving-average method (ARIMA), the non-linear auto-regressive neural networks (NAR), and the non-linear auto-regressive neural networks with exogenous inputs (NARX). Simple regression (empirical) models have been recalibrated or newly suggested in order to determine the improvement in prediction accuracy offered by the machine learning techniques. To assess the superiority of the new methods, different locations have been considered, including two stations in Cairo, Egypt, and nine other stations in five different countries in the MENA (Middle-East and North-Africa) region

Issued also as CD