DFG-NCN 2021/43/I/HS4/02578

PRobabilistic mid- and long-term prIce fORecasting In electriciTY markets (PRIORITY)

Grant no.: DFG-NCN 2021/43/I/HS4/02578 (DFG no. 505565850)
Funding agencies: German Research Foundation (DFG), Germany & National Science Centre (NCN), Poland
Funding scheme: OPUS LAP
Funding period: 1.09.2023-31.08.2026 (36 months)
Budget: 922 200 PLN + 416 000 EUR (ca. 2 807 300 PLN in total)
Title in Polish: Probabilistyczne prognozowanie cen na rynkach energii elektrycznej w horyzoncie średnio- i długoterminowym

Research team:

Principal Investigators (Kierownicy):

Investigators (Wykonawcy):

Collaborators (Współpracownicy):

– Ph.D. / M.Sc. / B.Sc. student

Aims and scope:

Climate change is urging fundamental transformations in almost all sectors of the economy. To reach the EU target of climate neutrality by 2050, the electricity system needs huge investments in new storage capacities, renewable energy generation, demand response systems and transmission grid infrastructure. However, such investments require a rigorous (in statistical terms: reliable and sharp) quantification of the uncertain future, especially of electricity prices for horizons ranging from months to years. Due to sector coupling, e.g., via electric vehicles and power-to-heat systems, the role of mid- and long-term probabilistic – or distributional – electricity price forecasting (EPF) will become even more important in the coming decades. Yet, the relevant literature is very scarce. Only 8% of EPF studies concern horizons beyond the day-ahead and less than 1% provide the so much needed midand long-term distributional forecasts.

It is exactly the aim of the PRIORITY project to address these challenges and develop methods and computational tools for PRobabilistic mid- and long-term prIce fORecasting In electriciTY markets. The project offers an integrated approach composed of four interrelated and performed in parallel work packages (WPs) and consisting of two complementary tasks each.

Tasks:

  • WP1: Machine learning forecasting models
    • WP1.1: Developing probabilistic deep neural networks and testing their applicability for mid- and long-term electricity price forecasting
    • WP1.2: Extending gradient boosting machines to cope with distributional electricity price forecasting beyond the day-ahead
  • WP2: Statistical and econometric forecasting models
    • WP2.1: Utilizing regularization to cope with high-dimensional models for probabilistic electricity price forecasting beyond the day-ahead
    • WP2.2: Constructing mid- and long-term regime-switching models with interpretable market regimes
  • WP3: Fundamental energy market forecasting models
    • WP3.1: Developing mid- and long-term fundamental power market models with stochastic inputs
    • WP3.2: Developing functional time series models for probabilistic forecasting of the supply and demand curves
  • WP4: Combination, reconciling and disaggregation
    • WP4.1: Developing probabilistic combination schemes for models with large temporal delays
    • WP4.2: Developing reconciling and disaggregation methods for mid- to longterm forecasting horizons

Publications:

Peer-reviewed articles in JCR-listed journals

2024 (1+), 2023 (0)

  1. A. Lipiecki, B. Uniejewski, R. Weron (2024) Postprocessing of point predictions for probabilistic forecasting of day-ahead electricity prices: The benefits of using isotonic distributional regression, Energy Economics 139, 107934  (doi: 10.1016/j.eneco.2024.107934). Earlier working paper version available from arXiv: https://arxiv.org/abs/2404.02270. Julia codes available from GitHub

Peer-reviewed articles in non JCR-listed journals

  1. …

Edited volumes / special issues

  1. …

Book chapters

  1. …

Conference papers

  1. T. Matsumoto, F. Ziel (2024) Electricity price forecasting with principal component-guided sparse regression, 20th International Conference on the European Energy Market (EEM24), 10.1109/EEM60825.2024.10609019

Forthcoming publications, submitted papers and work in progress

  1. P. Ghelasi, F. Ziel (2024) From day-ahead to mid and long-term horizons with econometric electricity price forecasting models, submitted. Working paper version available from arXiv: https://arxiv.org/abs/2406.00326
  2. M. Zimmermann, F. Ziel (2024) Efficient mid-term forecasting of hourly electricity load using generalized additive models, submitted. Working paper version available from arXiv: https://arxiv.org/abs/2405.17070
  3. M. Zimmermann, F. Ziel (2024) Spatial weather, socio-economic and political risks in probabilistic load forecasting, submitted. Working paper version available from arXiv: https://arxiv.org/abs/2408.00507

Developed software components

Project team meetings:

  • 2nd PRIORITY Meeting (part of the CrossFIT – PRIORITY joint meeting), 26.08.2024, Satellite to INREC 2024, 27-28.08.2024.
  • Kick-off meeting (part of the IMMORTAL – CrossFIT – PRIORITY joint meeting), 4.09.2023, Satellite to INREC 2023, 5-6.09.2023, Essen, Germany.