Assessing the Impact of Spatial Proximity Data on the Solar Insolation Prediction

Abstract

Improving the prediction of the availability of solar energy resources became a necessary component in the operation of utilities with a high penetration level of renewable energy resources. In this article, the solar insolation data in spatial proximity is leveraged to investigate the error in the prediction of solar insolation using multiple learning algorithms. Different error measures are utilized to evaluate the accuracy of the presented linear and nonlinear learning algorithms. Essential data preprocessing steps are conducted on the solar insolation data available from multiple meteorological stations in spatial proximity. The impact of utilizing the spatiotemporal data compared with the temporal data is analyzed. A comprehensive analysis based on multiple error measures is presented to compare the prediction error while employing multiple learning algorithms. It is shown that it is possible to identify the particular station and the particular learning algorithm that contribute the most in improving the solar insolation prediction of a specific location.