Investigation of using regression analysis and artificial neural network methods in estimate of solar radiation


DENİZ E. , Atik K.

ISI BILIMI VE TEKNIGI DERGISI-JOURNAL OF THERMAL SCIENCE AND TECHNOLOGY, cilt.27, ss.15-20, 2007 (SCI İndekslerine Giren Dergi)

  • Cilt numarası: 27 Konu: 2
  • Basım Tarihi: 2007
  • Dergi Adı: ISI BILIMI VE TEKNIGI DERGISI-JOURNAL OF THERMAL SCIENCE AND TECHNOLOGY
  • Sayfa Sayısı: ss.15-20

Özet

In this study, the more effective one of two methods, artificial neural network and regression analysis, was tried to be determined when they were used in estimation of the solar radiation intensity. For this purpose, wind velocity, air temperature, soil temperature, declination angle, humidity, the ratio of solar irradiation to daytime length, and monthly average of extraterrestrial solar radiation data between 1995 and 2004 belonging to Zonguldak city was obtained from Turkish State Meteorological Service (TSMS). Models were developed by regression analysis and artificial neural network (ANN) with the data obtained. Using these models, mountly average values of total solar irradiation between January/2005 and December/2005 were calculated and these calculated results were compared to measured results of the same period. It was determined that there are mean relative errors of 1.28 % and 3.25 % when the estimation was made by regression analysis and artificial neural network respectively.