A comparison between the application of empirical and ANN methods for estimation of daily global solar radiation in Iran

摘要: The present study generally aims to provide a comparison between the performance and suitability of different types models for estimation daily global solar radiation in Iran, based on duration sunshine hours diurnal air temperature. These consist empirical, ordinary ANN, ANN coupled with genetic algorithm (so called models). models’ was evaluated compared error statistics root mean squared (RMSE), bias (MBE), coefficient determination (R2). empirical (median R2, MBE, RMSE AP 0.93, 37.0, 179.3 J/cm2/day) could perform much better than MLP(n) 0.90, 55.7, 243.5 J/cm2/day). improved considerably after being by MLP-GA(n) 0.92, 38.4, 185.5 J/cm2/day), making them most accurate at stations studied. However, difference overall performances these ones slight. Lastly, despite had relatively accuracy ones, when taking metrics such as required processing time, skill, equipment into account, appear be suitable Iran.