Skies classification using artificial neural networks (ANN) techniques

In 2003, the International Commission on Illumination (CIE) adopted 15 standard skies that cover the whole probable spectrum of usual skies found in the world. Each sky represents a unique distribution. Once the standard sky has been identified, the sky irradiance and outdoor illuminance at any surfaces of interest can be obtained for subsequent investigations and complicated expressions for inclined surface models are not required. In Hong Kong where sky obstructions can be very large and of irregular shapes, sky distribution models that are specified by a given standard sky are more appropriate for such analyses. Long-term sky luminance data measurement is considered the most accurate approach of setting up the database. Alternatively, standard skies can be categorized by various climatic parameters of which the selection depends on their availability, accuracy, suitability and frequency of occurrence at a given location. Artificial neural networks (ANN) represent a powerful tool for pattern recognition. They learn the relationship between the input elements and the controllable and uncontrollable parameters by studying previous recorded data. This study presents the work on the standard sky classification using the ANN techniques. Solar and sky data recorded by our measuring station are used for the analysis. The findings showed that sky conditions can be correctly classified up to over 90% using the proposed approach.