Establishing new models for estimating the amounth of solar radiation using regression and multi-regression analysis

Before installing solar energy systems, the meteorological data of the area should be examined and solar potential should be considered at least for one year. Measurements are continuously recorded at meteorological stations. However, these measurements have difficulties in achieving cost-effective solar radiation measurement devices, qualification elements and uniform data collection. Due to technical failures, calibrations and other problems, it is impossible to collect data continuously and precisely. Because of this and other reasons, solar radiation prediction models for zones should be developed. In this study, daily and monthly average solar radiation models were developed using the data obtained from the Meteorology Department. Linear and multi-regression analyses have been used to improve these models. After these analyses Liner, Square, Cubic, Multi-1 and Multi-2 type models have been created. Daily and monthly average daily solar radiation forecasting models for Adana, Antakya, Antalya, Burdur, Elbistan, Isparta, Iskenderun, Karaman, Konya, Mersin, Nigde, Tarsus and Urgup have been gained to the literature. In order to see the predictive performance of these models, the evaluated results were compared with each other and with the results in the literature. In order to make this comparison, error analysis such as MBE, MBE%, RMSE, MAE, MAPE%, RSE and R² have been performed. The linear type regression model can sometimes be inadequate because of its dependence on only solar radiation and its duration. Therefore, the ability to estimate the addition of parameters such as relative humidity, temperature, soil temperature and cloudiness in a multi-type regression have been analysed. It has been observed that the models developed from the analysis results proved in many analyses that the cubic type models are better than the linear and quadratic type models in estimating the daily global radiation. In the estimation of the monthly mean solar radiation, the developed Multi-2 type model has been showed the best performance.