Hourly Global Solar Radiation Reconstruction Applying Machine Learning

Solar radiation significantly affected the cooling requirements of air-conditioned buildings during the summer period, attention is also paid to it in order to optimize the management of indoor lighting as it is a natural lighting source. However, the solar radiation is measured by a few weather stations in a few locations. The aim of this paper is to reconstruct the hourly solar global radiation trend on a year in some cities of the Northern and Southern Italy, starting from typical recorded meteorological data: temperature, relative humidity, wind speed and direction, etc. For this task a supervised machine learning algorithm has been used to build a model. The reached results show that the solar radiation hourly values can be extrapolated from other weather data in a reliable way, in fact an f-measure ranging from 0.950 an 1 is obtained for the several Italian cities involved in the experiment.