Naslov
Predicting Diffuse Horizontal Radiation with Neural Networks

Autori
Pavlinić, Bruno ; Tolić, Domagoj

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Skup
International Conference on Smart Systems and Technologies (SST)

Mjesto i datum
Osijek, Hrvatska, 19.10.2022. - 21.10.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
diffuse solar radiation ; neural networks ; prediction

Sažetak
Solar energy predictions are vital for low-cost evaluations of the long-term solar energy potential in a particular location. Solar radiation measurements acquired using various sensors are often deemed expensive, and therefore, affordable solutions for obtaining valid solar radiation data are sought after. There are three types of solar radiation (i.e., global, direct and diffuse), and each type requires different sensors and mechanisms for measuring it. If a solar station has sensors for measuring global solar radiation, it can use a Neural Network (NN) for predicting diffuse solar radiation without the need for expensive diffuse solar radiation sensors. Along with the global solar radiation, a station needs to have access to the zenith and azimuth angles, extraterrestrial radiation on a horizontal surface, extraterrestrial radiation normal to the sun, precipitable water, and aerosol optical depth. Using all those measurements, we predict diffuse solar radiation with high accuracy, which allows solar stations to have diffuse radiation data without oftentimes prohibitive costs.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo

POVEZANOST RADA