RNN based solar radiation forecasting using adaptive learning rate

Abstract

The estimation of solar irradiation data is very important for renewable energy and solar energy systems applications. The forecasts can be used to predict the output power of photovoltaic systems installed in power systems and control the output of other generators to meet the electricity demand. In this paper, a Recurrent Neural Network(RNN) model is used to forecast the Daily, Mean Monthly and Hourly Solar Irradiations using the recorded meteorological data. Here, an adaptive learning rate is proposed for the RNN. The results of the RNN is compared with that of a Multi Layer perceptron(MLP). It is found that the RNN with the adaptive learning rate gives a better performance than the conventional feed forward network. © 2013 Springer International Publishing.