Main Article Content

Dr.A.Immanuel Selvakumar
R.Meenal R.Meenal

Abstract

Solar radiation is the essential parameter required for all solar energy applications. In India, this radiation data is not available in all the locations due to the higher cost of measuring equipment and techniques involved. In this work, Genetic algorithm (GA) technique based global solar radiation (GSR) prediction is proposed. To achieve this, Angstrom model based on sunshine duration is employed. Generally sunshine-based models are very accurate. However sunshine records are not available in all the locations. For this purpose, temperature-based regression model is also developed to estimate the GSR. The regression coefficients for the two models are separately determined using the conventional statistical regression technique (SRT) and GA. The monthly average daily GSR on horizontal surface is estimated for three different regions of India namely Bhubaneswar, Nagpur and New Delhi using these techniques. The estimated GSR values are compared with measured GSR data of India Meteorological Department (IMD), Pune. The comparisons are based on statistical parameters namely MBE, RMSE, MAPE and Correlation coefficient R. In addition, the t-statistics is used to determine the statistical significance of the model estimates. The results show that the regression coefficients derived from the proposed GA, performs better than the conventional SRT in estimating the GSR in India. Hence GA technique can be effective in estimating solar radiation wherever the radiation data is not avai

Article Details