INTERNATIONAL ENERGY JOURNALVolume 20, No. 02, Month JUNE, Year 2020, Pages 181 - 200
Stochastic model for generating synthetic hourly global horizontal solar radiation data sets based on auto regression characterization
M.P. Anand, Athula Rajapakse, Bagen Bagen
Abstract Download PDFA large number of non-repetitive multi-year hourly solar radiation time-series dataset are desired when applying Monte Carlo techniques for planning and design of solar energy systems. Solar radiation models which utilize the clearness index and average value decomposition methods are commonly used to generate synthetic set of solar irradiances for this purpose. In this paper, a novel stochastic solar radiation model based on probability distributions of the first-order differences of hourly global solar horizontal radiation is proposed. The first-order differences are modeled using a trend component and a stochastic component represented using the cumulative distribution functions, both extracted from historical data taken over a window of 31 days around the considered day of the year. Measured solar radiation data from four different locations with varying climate characteristics were used to evaluate the proposed model in comparison to two previously reported models. The proposed method performed consistently better in terms of the similarity of probability distributions and autocorrelation functions, for all four locations and datasets
first order differences of solar radiation, long-term solar radiation models, Monte Carlo simulation, solar irradiance, synthetic hourly solar radiation.