Assessment and mapping of solar energy potential using artificial neural network and GIS technology in the southern part of India

Abstract

Prediction and assessment of solar radiation are necessary pre-requisites in developing solar technology. Here, an artificial neural network (ANN) model has been developed to predict solar energy potential in the Southern part of India: Andhra Pradesh (AP) and Telangana State (TS), lie between 12°41' and 22°N latitude and 77° and 84°40'E longitude. Generalized feed-forward with back-propagation neural networks were considered using MATLAB. Three layered neural network with different architectures are designed and evaluated. For training and testing the network, geographical and meteorological data of 28 sites over a period of recent 22 years from the NASA geo-satellite database were taken. Geographical parameters (latitude, longitude and altitude), meteorological data (temperature, sunshine duration, relative humidity and precipitation) were used as input data, whereas the mean solar radiation was used as the output of the network. All the parameters taken here are in the form of monthly mean. The ANN model has been evaluated for test locations by calculating mean absolute percentage error (MAPE). The correlation coefficients (R-value) between the output of model and the measured value of solar radiation is calculated.  The R-value were more than 0.95, which show high reliability of the model for prediction of solar radiation anywhere within AP and TS. Solar radiation of major cities was predicted using developed model. Predicted solar radiation is analyzed and used to create monthly mean maps using GIS technology. These maps can be useful to estimate solar energy potential at any locations within AP and TS.