DEVELOPMENT OF A NEURAL NETWORK MODEL FOR SOLAR RADIATION ESTIMATION IN KANO STATE NIGERIA

Abstract

In this paper, the global solar radiation on the horizontal surface in Kano, Nigeria using 10-year data ranging from (2002-2012) was analysed based on the series of measured meteorological data: monthly mean (minimum temperature, maximum temperature, sunshine hours and relative humidity). Artificial Neural Network (ANN) was used to model the solar radiation. The ANN was fed with four input layers, monthly mean (minimum temperature, maximum temperature, sunshine hour and relative humidity) and one output layer (monthly mean solar radiation). 120 sets of data were used, 96 sets (80%) for training and 24 sets (20%) for testing. The root mean square error (RMSE), correlation coefficient (R), were used to assess the model performance. The results obtained indicates a good correlation between the measured and estimated result, the statistical parameters for the training phase are RSME= 0.9505MJ/m² and R= 0.92995MJ/m². This proves ANN soft computing technique as a good tool for solar radiation prediction.