A class-driven approach based on long short-term memory networks for electricity price scenario generation and reduction

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Uncertainty characterization is an essential component of decision-making problems in electricity markets. In this work, a class-driven approach is proposed to describe stochasticity. The methodology consists of a three-step process that includes a class allocation component, a generative element based on a long short-term memory neural network and an automated reduction method with a variance-based continuation criterion. The system is employed and evaluated on Dutch imbalance market prices. Test results are presented, expressing the proficiency of the approach, both in generating realistic scenario sets that reflect the erratic dynamics in the data and adequately reducing generated sets without the need to explicitly and manually predetermine the cardinality of the reduced set.
Original languageEnglish
JournalIEEE Transactions on Power Systems
DOIs
Publication statusAccepted/In press - 13 Jan 2020

Fingerprint

Electricity
Decision making
Neural networks
Uncertainty
Power markets
Long short-term memory

Keywords

  • Artificial intelligence
  • Deep learning
  • Long short-term memory (lstm)
  • Recurrent neural networks (rnn)
  • Uncertainty
  • Machine learning
  • Scenario generation
  • Scenario reduction
  • Imbalance prices

Cite this

@article{a016d41a834e4bb3906bfd421a0a8b8d,
title = "A class-driven approach based on long short-term memory networks for electricity price scenario generation and reduction",
abstract = "Uncertainty characterization is an essential component of decision-making problems in electricity markets. In this work, a class-driven approach is proposed to describe stochasticity. The methodology consists of a three-step process that includes a class allocation component, a generative element based on a long short-term memory neural network and an automated reduction method with a variance-based continuation criterion. The system is employed and evaluated on Dutch imbalance market prices. Test results are presented, expressing the proficiency of the approach, both in generating realistic scenario sets that reflect the erratic dynamics in the data and adequately reducing generated sets without the need to explicitly and manually predetermine the cardinality of the reduced set.",
keywords = "Artificial intelligence, Deep learning, Long short-term memory (lstm), Recurrent neural networks (rnn), Uncertainty, Machine learning, Scenario generation, Scenario reduction, Imbalance prices",
author = "Bart Stappers and N.G. Paterakis and Kok, {J.K. (Koen)} and M. Gibescu",
year = "2020",
month = "1",
day = "13",
doi = "10.1109/TPWRS.2020.2965922",
language = "English",
journal = "IEEE Transactions on Power Systems",
issn = "0885-8950",
publisher = "Institute of Electrical and Electronics Engineers",

}

TY - JOUR

T1 - A class-driven approach based on long short-term memory networks for electricity price scenario generation and reduction

AU - Stappers, Bart

AU - Paterakis, N.G.

AU - Kok, J.K. (Koen)

AU - Gibescu, M.

PY - 2020/1/13

Y1 - 2020/1/13

N2 - Uncertainty characterization is an essential component of decision-making problems in electricity markets. In this work, a class-driven approach is proposed to describe stochasticity. The methodology consists of a three-step process that includes a class allocation component, a generative element based on a long short-term memory neural network and an automated reduction method with a variance-based continuation criterion. The system is employed and evaluated on Dutch imbalance market prices. Test results are presented, expressing the proficiency of the approach, both in generating realistic scenario sets that reflect the erratic dynamics in the data and adequately reducing generated sets without the need to explicitly and manually predetermine the cardinality of the reduced set.

AB - Uncertainty characterization is an essential component of decision-making problems in electricity markets. In this work, a class-driven approach is proposed to describe stochasticity. The methodology consists of a three-step process that includes a class allocation component, a generative element based on a long short-term memory neural network and an automated reduction method with a variance-based continuation criterion. The system is employed and evaluated on Dutch imbalance market prices. Test results are presented, expressing the proficiency of the approach, both in generating realistic scenario sets that reflect the erratic dynamics in the data and adequately reducing generated sets without the need to explicitly and manually predetermine the cardinality of the reduced set.

KW - Artificial intelligence

KW - Deep learning

KW - Long short-term memory (lstm)

KW - Recurrent neural networks (rnn)

KW - Uncertainty

KW - Machine learning

KW - Scenario generation

KW - Scenario reduction

KW - Imbalance prices

U2 - 10.1109/TPWRS.2020.2965922

DO - 10.1109/TPWRS.2020.2965922

M3 - Article

JO - IEEE Transactions on Power Systems

JF - IEEE Transactions on Power Systems

SN - 0885-8950

ER -