Uncertainty reduction in power generation forecast using coupled wavelet-ARIMA

Abstract

Forecasts errors in variable generation and load are usually assumed to be normally distributed. This supposition is used as the basis for estimating these errors' uncertainty and consequences for a power system. Another assumption is that the forecast errors are stationary processes with time-independent probability distributions. These hypotheses, however, are not always valid. In this paper, we introduce a new approach without implying normal distributions and stationarity of forecast errors. In addition, it is desired to more accurately quantify the forecast uncertainty by reducing prediction intervals of forecasts. We use automatically coupled wavelet transform and autoregressive integrated moving-average (ARIMA) forecasting to reflect multi-scale variability of forecast errors. The proposed analysis reveals slow-changing “quasi-deterministic” components of forecast errors. This helps improve forecasts produced by other means, e.g., using weather-based models, and reduce forecast errors prediction intervals. Thus, we demonstrate the possibility of uncertainty reduction for wind, solar, and load forecast errors by 10–12%.