ABSTRACT

This study focuses on the development of connectionist model such as neural network based method to efficiently predict solar radiation of a particular place. Here a comparative study is given between a conventional approach, that is Multiple Linear Regression (MLR) with the soft computing approach, that is Multi Layer Perceptron Neural Network (MLP).In each approach the input parameters are taken as several meteorological parameters that influences solar irradiance . The estimated values of solar radiation are compared with measured values in terms of percentage error and Root Mean Square Error (RMSE).To improve prediction accuracy Feature Selection technique is used which can select good features on-line while learning the prediction task. The paper also gives a comparative study of results of using both MLP and FSMLP.