Solar generation prediction using the ARMA model in a laboratory-level micro-grid

• Published in: Smart Grid Communications pp. 528 - 533
• Main Authors: Rui Huang, Tiana Huang, Gadh, R., Na Li
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2012
• Subjects: Data models; Forecasting; Laboratories; Mathematical model; Predictive models; Satellites; Solar radiation
• ISBN: 9781467309103; 1467309109
• DOI: 10.1109/SmartGridComm.2012.6486039

The goal of this article is to investigate and research solar generation forecasting in a laboratory-level micro-grid, using the UCLA Smart Grid Energy Research Center (SMERC) as the test platform. The article presents an overview of the existing solar forecasting models and provides an evaluation of various solar forecasting providers. The auto-regressive moving average (ARMA) model and the persistence model are used to predict the future solar generation within the vicinity of UCLA. In the forecasting procedures, the historical solar radiation data originates from SolarAnywhere. System Advisor Model (SAM) is applied to obtain the historical solar generation data, with inputting the data from SolarAnywhere. In order to validate the solar forecasting models, simulations in the System Identification Toolbox, Matlab platform are performed. The forecasting results with error analysis indicate that the ARMA model excels at short and medium term solar forecasting, whereas the persistence model performs well only under very short duration.