Day-ahead solar forecasting with endogenous data

What is it about?

Renewable energy forecasting services comprise various modules for intra-day and day-ahead forecasts. This work specifically addresses day-ahead forecasts, utilizing specifications based on endogenous, historical measurements. These specifications are designed to be computationally efficient, requiring fewer input variables and less training data. Such weather-independent specifications serve as benchmarks against the more computationally demanding forecasts based on numerical weather predictions.

Featured Image

Why is it important?

Our work assesses methodologies that leverage endogenous historical measurements for day-ahead forecasting of solar irradiance and energy yield. The wide array of models evaluated includes variants of (i) functional time series specifications, featuring two innovative procedures; (ii) time series nearest neighbor (NN) approaches; (iii) exponential smoothing procedures, which have shown excellent performance in previous forecasting competitions;10 (iv) autoregressive integrated moving average (ARIMA) models, serving as a standard benchmark;11 (v) automatic time series decomposition-based techniques, anticipated to excel in series with pronounced periodicity; and (vi) the persistence (PRS) model. Each model category offers viable components for forecasting systems, characterized by their efficiency in computation, minimal requirement for input variables (thus reducing associated costs), and short training periods.