Non-Stationarity in Multiagent Reinforcement Learning in Electricity Market Simulation

Charles Renshaw-Whitman; Viktor Zobernig; Jochen Cremer; Laurens DE Vries

Non-Stationarity in Multiagent Reinforcement Learning in Electricity Market Simulation

Charles Renshaw-Whitman, Viktor Zobernig (Vortragende:r), Jochen Cremer, Laurens DE Vries

Integrated Energy Systems

Delft University of Technology

Publikation: Beitrag in Buch oder Tagungsband › Vortrag mit Beitrag in Tagungsband › Begutachtung

2 Downloads (Pure)

Abstract

The design of electricity markets may be facilitated by simulating actors’ behaviors. Recent studies model human decision-makers within markets as agents which learn strategies that maximize expected profits. This work investigates the problem
of ’non-stationarity’ in the context of market simulations, a problem with the learning-algorithms used by such studies which results in agents behaving irrationally, thus limiting the studies’ applicability to real-world strategic behavior.

Originalsprache	Englisch
Titel	23rd Power Systems Computation Conference
Seiten	1-9
Seitenumfang	9
Publikationsstatus	Veröffentlicht - 4 Juni 2024
Veranstaltung	23rd Power Systems Computation Conference - Paris, France, Paris, Frankreich Dauer: 4 Juni 2024 → 7 Juni 2024 Konferenznummer: 23 https://pscc2024.fr/

Konferenz

Konferenz	23rd Power Systems Computation Conference
Kurztitel	PSCC 2024
Land/Gebiet	Frankreich
Stadt	Paris
Zeitraum	4/06/24 → 7/06/24
Internetadresse	https://pscc2024.fr/

Research Field

Flexibility and Business Models

Zugriff auf Dokument

2024_187 Non-stationarity in Multi-Agent ReinforcementEndgültige, publizierte Fassung, 435 KBLizenz: Nicht angegeben

Andere Dateien und Links

https://pscc2024.fr/

Diese Publikation zitieren

@inproceedings{77b8844db361401bb6f2594ea45671a6,

title = "Non-Stationarity in Multiagent Reinforcement Learning in Electricity Market Simulation",

abstract = "The design of electricity markets may be facilitated by simulating actors{\textquoteright} behaviors. Recent studies model human decision-makers within markets as agents which learn strategies that maximize expected profits. This work investigates the problemof {\textquoteright}non-stationarity{\textquoteright} in the context of market simulations, a problem with the learning-algorithms used by such studies which results in agents behaving irrationally, thus limiting the studies{\textquoteright} applicability to real-world strategic behavior.",

keywords = "Deep learning, Game theory, Market design, Market Simulation, Reinforcement Learning",

author = "Charles Renshaw-Whitman and Viktor Zobernig and Jochen Cremer and {DE Vries}, Laurens",

note = "Paper presented at the 23rd Power Systems Computation Conference (PSCC 2024) Paris, France, June 4-7, 2024; 23rd Power Systems Computation Conference, PSCC 2024 ; Conference date: 04-06-2024 Through 07-06-2024",

year = "2024",

month = jun,

day = "4",

language = "English",

pages = "1--9",

booktitle = "23rd Power Systems Computation Conference",

url = "https://pscc2024.fr/",

}

TY - GEN

T1 - Non-Stationarity in Multiagent Reinforcement Learning in Electricity Market Simulation

AU - Renshaw-Whitman, Charles

AU - Cremer, Jochen

AU - DE Vries, Laurens

A2 - Zobernig, Viktor

N1 - Conference code: 23

PY - 2024/6/4

Y1 - 2024/6/4

N2 - The design of electricity markets may be facilitated by simulating actors’ behaviors. Recent studies model human decision-makers within markets as agents which learn strategies that maximize expected profits. This work investigates the problemof ’non-stationarity’ in the context of market simulations, a problem with the learning-algorithms used by such studies which results in agents behaving irrationally, thus limiting the studies’ applicability to real-world strategic behavior.

AB - The design of electricity markets may be facilitated by simulating actors’ behaviors. Recent studies model human decision-makers within markets as agents which learn strategies that maximize expected profits. This work investigates the problemof ’non-stationarity’ in the context of market simulations, a problem with the learning-algorithms used by such studies which results in agents behaving irrationally, thus limiting the studies’ applicability to real-world strategic behavior.

KW - Deep learning

KW - Game theory

KW - Market design

KW - Market Simulation

KW - Reinforcement Learning

UR - https://pscc2024.fr/

M3 - Conference Proceedings with Oral Presentation

SP - 1

EP - 9

BT - 23rd Power Systems Computation Conference

T2 - 23rd Power Systems Computation Conference

Y2 - 4 June 2024 through 7 June 2024

ER -