Abstract
Machine learning (ML) models were developed to estimate monthly average diffuse solar radiation (DSR) with sky-clearness index and relative sunshine period as inputs for the humid subtropical climate of India. Three categories of ML models were defined, each having six models. The solar radiation data was split into two parts, the “Training dataset” used to develop the models and the “Validation dataset” used to test the models. Model accuracy was investigated as a function of various commonly used statistical pointers. A comparison was made between the ML models and well-known empirical models from prior research. Global performance indicator was used to rank ML models within each category. The projected values from the ML models and solar radiation data were in reasonable agreement. Thus, DSR can be accurately predicted using ML models in the area under consideration.
Original language | English |
---|---|
Article number | e13973 |
Journal | Environmental Progress and Sustainable Energy |
Volume | 42 |
Issue number | 2 |
DOIs | |
Publication status | Published - 1 Mar 2023 |
Externally published | Yes |
Keywords
- artificial neural networks
- solar energy
- sustainability