Predicting Solar Irradiance Intervals for Power Generation Systems

To address the inadequacy in interval prediction accuracy of solar irradiance in photovoltaic power generation systems, this study establishes a predictive model based on Variational Mode Decomposition (VMD) and Autoregressive Integrated Moving Average Model (ARIMA). The model achieves the prediction of the overall distribution of solar irradiance intervals under unknown sample conditions. First, to overcome the non-stationarity issue in the high-complexity and strongly nonlinear solar irradiance time series, the Variational Mode Decomposition method is employed to decompose the original solar irradiance time series. Next, using the ARIMA model, point predictions for each component are derived. Finally, employing non-parametric Gaussian Kernel Density Estimation, the error between the point prediction results and the original data is estimated, yielding the interval prediction results for solar irradiance. Experimental results demonstrate that, compared to other models, this interval prediction model better addresses the bias in prediction results caused by the fluctuation and randomness of solar irradiance.