Clearness index predicting using an integrated artificial neural network (ANN) approach

Abstract

Accurate insolation data for many cities and locations is not available. Therefore estimation of such data for solar applications is inevitable. Clearness index KT¯ is one of the parameters which represent the atmosphere characteristics and solar energy potential at a location. This paper presents an Integrated Artificial Neural Network (ANN) approach for optimum forecasting of Clearness index by considering environmental and meteorological factors. The ANN train and test data with multi-layer perceptron (MLP) approach which is popular and applicable network for such engineering investigations is used in this study. The proposed approach is particularly useful for locations with no available measurement equipment. To show the applicability and superiority of the integrated ANN approach, monthly data were collected for 30 years (1975–2005) in 19 nominal cities in Iran. The acquired results of the model have shown high accuracy with a mean absolute percentage error (MAPE) about 4.338%. Furthermore, a detailed analysis is performed on the various combinations of input parameters. Finally, using these results, geographic information system (GIS) map is produced and presented. This map is very good indicative of climate and solar potential of different locations based on ANN analysis.