As the demand for renewable energy is increasing for power generation, the penetration of solar energy is also on the rise. However, as solar energy is highly uncertain, accurate prediction of solar energy is crucial for performing necessary calculations and maintaining stability. To enhance the accuracy and stability of the system, Machine Learning (ML) methods for forecasting have been widely employed. In this study, four ML methods-random forest regression, support vector, multilinear regression, and decision tree are utilized and compared based on metrics such as root mean square error, mean square error, mean absolute error, and R2 for three case studies. Initially, errors are calculated without hyperparameter tuning. Subsequently, the same methods are executed with hyperparameter tuning, and it is observed that hyperparameter tuning helps minimize errors. The case study is also conducted for different seasons. For this research, data is sourced from the National Solar Radiance Database over a two-year period from 2018 to 2019.
Impact of Hyperparameter Tunning in Hourly Solar Irradiance Forecasting by Using Machine Learning Techniques
2024-07-31
1425126 byte
Conference paper
Electronic Resource
English
Airport rail services: forecasting hourly demand
British Library Conference Proceedings | 2007
|Forecasting of solar irradiance with just‐in‐time modeling
Online Contents | 2013
|