On cointegration for modeling and forecasting wind power production

TL;DR

This paper investigates the use of cointegrated VAR models to enhance short-term wind power forecasting accuracy by leveraging potential long-term relationships between multiple wind power time series.

Contribution

It demonstrates that incorporating cointegration into VAR models can improve short-term wind power forecasts, based on empirical analysis of German wind data.

Findings

Cointegration improves forecast accuracy for short-term wind power prediction.

Accounting for long-term relationships enhances model performance.

Empirical results show benefits for German wind data.

Abstract

This study evaluates the performance of cointegrated vector autoregressive (VAR) models for very short- and short-term wind power forecasting. Preliminary results for a German data set comprising six wind power production time series indicate that taking into account potential cointegrating relations between the individual series can improve forecasts at short-term time horizons.