Entwicklung eines Verfahrens zur Strompreisvorhersage im kurzfristigen Intraday-Handelszeitraum

  • Forecasting method for intraday power prices leading to market closure

Bader, Andreas; Haubrich, Hans-Jürgen (Thesis advisor); Sillaber, Alfons (Thesis advisor); de Doncker, Rik W. (Thesis advisor)

1. Auflage. - Aachen : E.ON Energy Research Center, RWTH Aachen University (2017)
Book, Dissertation / PhD Thesis

In: E.ON Energy Research Center 45. Ausgabe : PGS, Power Generation and Storage Systems
Page(s)/Article-Nr.: 1 Online-Ressource (ix, 115 Seiten) : Diagramme

Dissertation, RWTH Aachen University, 2017


A complete power price forecast, spanning various time horizons from intraday to long-term, is commercially and economically beneficial. Although the liquidity of intraday markets has grown substantially over the past few years, the main focus of research still lies in long-term price forecasting. However, improved modelling of intraday prices enables market players to further optimize their renewable portfolios within day and to generate additional revenues with flexible units. This thesis presents a newly developed method of forecasting intraday power prices for the last trading hours before market closure based on fundamental influencing factors. The method combines mainly Markov chains and regression analysis to predict intraday price movements and includes various parameters which can be modified to model different price scenarios. In order to model the price movements a new continuous time-dependent intraday index is defined and introduced in this thesis.Markov chains are applied to intraday prices to depict stochastic historic price movements. The resulting probability distributions are found to be dependent on the price level and the lead-time to delivery. The Gaussian nature of the distributions indicates that consideration of fundamental factors is necessary to derive the direction of price development. Therefore, regression analysis is applied to model fundamental influencing factors, such as activated control reserve, intraday deviations of renewable production forecasts and demand forecast errors. The highest impact on intraday prices results from the activation of control reserve, which is positively correlated to the prices. Furthermore, the model shows a clear negative correlation between price movements and intraday deviations of wind and solar production forecasts. An impact based on demand forecast errors is not ascertainable. A considerable improvement of the modelling results can be achieved through a combination of multiple fundamental influencing factors. Furthermore, the impact of various model parameters on the intraday price index is shown, such as a variation of the forecast lead-time to the trading period.