Abstract
The percentage of renewable energies (RE) within power generation in Germany has increased significantly since 2010 from 16.6% to 42.9% in 2019 which led to a larger variability in the electricity prices. In particular, generation from wind and photovoltaics induces high volatility, is difficult to forecast and challenging to plan. To counter this variability, the continuous intraday market at the EPEX SPOT offers the possibility to trade energy in a short-term perspective, and enables the adjustment of earlier trading errors. In this context, appropriate price forecasts are important to improve the trading decisions on the energy market. Therefore, we present and analyse in this paper a novel approach for the prediction of the energy price for the continuous intraday market at the EPEX SPOT. To model the continuous intraday price, we introduce a semi-continuous framework based on a rolling window approach. For the prediction task we utilise shallow learning techniques and present a LSTM-based deep learning architecture. All approaches are compared against two baseline methods which are simply current intraday prices at different aggregation levels. We show that our novel approaches significantly outperform the considered baseline models. In addition to the general results, we further present an extension in form of a multi-step ahead forecast.
C. Scholz and M. Lehna—Both authors contributed equally to this work.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
- 2.
Note that it is planned for future work to consider further training and testing periods to investigate the quality of the models.
- 3.
Note that two days were excluded from the observation (01.01.2018 & 25.03.2018). The first was excluded due to missing training data of 2017, while the second was generally missing data at the respective day.
- 4.
- 5.
While many researchers further transform the spot price, e.g. Uniejewski and Weron [19], we were not able to detect improvements in our estimation. Instead, we opted for the simple scaling through a constant c so that \(99.7 \%\) of the data was within the interval \([-1,1]\).
- 6.
- 7.
At the time of writing, we had no adequate photovoltaic feed-in forecast available.
- 8.
- 9.
- 10.
For regression tasks typically m/6 features are selected at random.
- 11.
- 12.
As example, for the prediction interval 20:00-20:15 the price of 19.59 is taken as \(BL_{1}\)baseline and the volume weighted mean of 19:45-20.00 as \(BL_{15}\).
- 13.
Since the RF had similar values to the XGB we skip the analysis of the RF.
- 14.
- 15.
The three input gates are kept in the original structure.
References
Andrade, J.R., Filipe, J., Reis, M., Bessa, R.J.: Probabilistic Price Forecasting for Day-Ahead and Intraday Markets: Beyond the Statistical Model (2017)
Breiman, L.: Random forests. Mach. Learn. 45, 5–32 (2001)
Chen, T., Guestrin, C.: XGBoost: a scalable tree boosting system. In: Proceedings of the 22nd International Conference on Knowledge Discovery and Data Mining, pp. 785–794 (2016)
Frade, P., Vieira-Costa, J.V., Osório, G.J., Santana, J.J., Catalão, J.P.: Influence of wind power on intraday electricity spot market: a comparative study based on real data. Energies 11, 2974 (2018)
Guo, C., Berkhahn, F.: Entity Embeddings of Categorical Variables. arXiv (2016)
Gürtler, M., Paulsen, T.: Forecasting performance of time series models on electricity spot markets: a quasi-meta-analysis. Int. J. Energy Sect. Manage. 12, 103–129 (2018)
Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9, 1735–1780 (1997)
Kath, C.: Modeling intraday markets under the new advances of the cross-border intraday project (XBID): evidence from the german intraday market. Energies 12, 4339 (2019)
Kath, C., Ziel, F.: The value of forecasts: quantifying the economic gains of accurate quarter-hourly electricity price forecasts. Energy Econ. 76, 411–423 (2018)
Kiesel, R., Paraschiv, F.: Econometric analysis of 15-minute intraday electricity prices. Energy Econ. 64, 77–90 (2017)
Li, L., Jamieson, K., DeSalvo, G., Rostamizadeh, A., Talwalkar, A.: Hyperband: a novel bandit-based approach to hyperparameter optimization. J. Mach. Learn. Res. 18, 6765–6816 (2017)
Martin, H.: A Limit Order Book Model for the German lntraday Electricity Market
Monteiro, C., Ramirez-Rosado, I.J., Fernandez-Jimenez, L.A., Conde, P.: Short-term price forecasting models based on artificial neural networks for intraday sessions in the iberian electricity market. Energies 9, 721 (2016)
Narajewski, M., Ziel, F.: Econometric Modelling and Forecasting of Intraday Electricity Prices. Journal of Commodity Markets p. 100107 (2019)
Oksuz, I., Ugurlu, U.: Neural network based model comparison for intraday electricity price forecasting. Energies 12, 4557 (2019)
Pape, C., Hagemann, S., Weber, C.: Are fundamentals enough? explaining price variations in the german day-ahead and intraday power market. Energy Econ. 54, 376–387 (2016)
Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Learning representations by back-propagating errors. Nature 323, 533–536 (1986)
Uniejewski, B., Marcjasz, G., Weron, R.: Understanding intraday electricity markets: variable selection and very short-term price forecasting using LASSO. Int. J. Forecast. 35, 1533–1547 (2019)
Uniejewski, B., Weron, R.: Efficient forecasting of electricity spot prices with expert and lasso models. Energies 11, 2039 (2018)
Wessel, A., Dobschinski, J., Lange, B.: Integration of offsite wind speed measurements in shortest-term wind power prediction systems. In: 8th International Workshop on Large-Scale Integration of Wind Power into Power Systems, pp. 14–15 (2009)
Ziel, F.: Modeling the impact of wind and solar power forecasting errors on intraday electricity prices. In: 2017 14th International Conference on the European Energy Market (EEM), pp. 1–5 (2017)
Acknowledgement
This work was supported as Fraunhofer Cluster of Excellence Integrated Energy Systems CINES.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
Appendix a Orderbook Data
(See Table 3)
Appendix B LSTM Architecture
Structure of the LSTM model. On the left side, the three embedding layers process the categorical input. In the middle branch, the look-back variables of the last 4h are feed into the LSTMlayer. On the right side, the last 15 min are also implemented into the network.
Appendix C Model Hyper Parameters
In this section we display the hyper parameters for the RF in Table 4, the XGB in Table 5 and the LSTM in Table 6.
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Scholz, C., Lehna, M., Brauns, K., Baier, A. (2021). Towards the Prediction of Electricity Prices at the Intraday Market Using Shallow and Deep-Learning Methods. In: Bitetta, V., Bordino, I., Ferretti, A., Gullo, F., Ponti, G., Severini, L. (eds) Mining Data for Financial Applications. MIDAS 2020. Lecture Notes in Computer Science(), vol 12591. Springer, Cham. https://doi.org/10.1007/978-3-030-66981-2_9
Download citation
DOI: https://doi.org/10.1007/978-3-030-66981-2_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-66980-5
Online ISBN: 978-3-030-66981-2
eBook Packages: Computer ScienceComputer Science (R0)
