Demand Response through Price-setting Multi-agent Reinforcement Learning

Morten Herget Christensen; Cédric Ernewein; Pierre Pinson

doi:10.1145/3427773.3427862

Demand Response through Price-setting Multi-agent Reinforcement Learning

Morten Herget Christensen
, Cédric Ernewein
, Pierre Pinson

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

Price based demand response is a cost-effective way of obtaining flexibility needed in power systems with high penetration of intermittent renewable energy sources. Model-free deep reinforcement learning is proposed as a way to train autonomous agents for enabling buildings to participate in demand response programs as well as coordinating such programs though price setting in a multiagent setup. First, we show price responsive control of buildings with electric heat pumps using deep deterministic policy gradient. Then a coordinating agent is trained to manage a population of buildings by adjusting the price in order to keep the total load from exceeding the available capacity considering also the non-flexible base load.

Original language	English
Title of host publication	RLEM'20: Proceedings of the 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities
Number of pages	5
Publisher	Association for Computing Machinery
Publication date	2020
ISBN (Print)	9781450381932
DOIs	https://doi.org/10.1145/3427773.3427862
Publication status	Published - 2020
Event	1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities - Online Duration: 17 Nov 2020 → 17 Nov 2020 Conference number: 1 https://rlem-workshop.net/

Workshop

Workshop	1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities
Number	1
Location	Online
Period	17/11/2020 → 17/11/2020
Internet address	https://rlem-workshop.net/

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Access to Document

10.1145/3427773.3427862

OpenUrl availability

Check availability in DTU Findit

Cite this

@inproceedings{078bcf26c324448a9d2832bb7bfb9f54,

title = "Demand Response through Price-setting Multi-agent Reinforcement Learning",

abstract = "Price based demand response is a cost-effective way of obtaining flexibility needed in power systems with high penetration of intermittent renewable energy sources. Model-free deep reinforcement learning is proposed as a way to train autonomous agents for enabling buildings to participate in demand response programs as well as coordinating such programs though price setting in a multiagent setup. First, we show price responsive control of buildings with electric heat pumps using deep deterministic policy gradient. Then a coordinating agent is trained to manage a population of buildings by adjusting the price in order to keep the total load from exceeding the available capacity considering also the non-flexible base load.",

author = "Christensen, \{Morten Herget\} and C{\'e}dric Ernewein and Pierre Pinson",

year = "2020",

doi = "10.1145/3427773.3427862",

language = "English",

isbn = "9781450381932",

booktitle = "RLEM'20: Proceedings of the 1st International Workshop on Reinforcement Learning for Energy Management in Buildings \& Cities",

publisher = "Association for Computing Machinery",

address = "United States",

note = "1st International Workshop on Reinforcement Learning for Energy Management in Buildings \& Cities, RLEM'20 ; Conference date: 17-11-2020 Through 17-11-2020",

url = "https://rlem-workshop.net/",

}

Christensen, MH, Ernewein, C & Pinson, P 2020, Demand Response through Price-setting Multi-agent Reinforcement Learning. in RLEM'20: Proceedings of the 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities. Association for Computing Machinery, 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities, 17/11/2020. https://doi.org/10.1145/3427773.3427862

Demand Response through Price-setting Multi-agent Reinforcement Learning. / Christensen, Morten Herget; Ernewein, Cédric; Pinson, Pierre.
RLEM'20: Proceedings of the 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities. Association for Computing Machinery, 2020.

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - Demand Response through Price-setting Multi-agent Reinforcement Learning

AU - Christensen, Morten Herget

AU - Ernewein, Cédric

AU - Pinson, Pierre

N1 - Conference code: 1

PY - 2020

Y1 - 2020

N2 - Price based demand response is a cost-effective way of obtaining flexibility needed in power systems with high penetration of intermittent renewable energy sources. Model-free deep reinforcement learning is proposed as a way to train autonomous agents for enabling buildings to participate in demand response programs as well as coordinating such programs though price setting in a multiagent setup. First, we show price responsive control of buildings with electric heat pumps using deep deterministic policy gradient. Then a coordinating agent is trained to manage a population of buildings by adjusting the price in order to keep the total load from exceeding the available capacity considering also the non-flexible base load.

AB - Price based demand response is a cost-effective way of obtaining flexibility needed in power systems with high penetration of intermittent renewable energy sources. Model-free deep reinforcement learning is proposed as a way to train autonomous agents for enabling buildings to participate in demand response programs as well as coordinating such programs though price setting in a multiagent setup. First, we show price responsive control of buildings with electric heat pumps using deep deterministic policy gradient. Then a coordinating agent is trained to manage a population of buildings by adjusting the price in order to keep the total load from exceeding the available capacity considering also the non-flexible base load.

U2 - 10.1145/3427773.3427862

DO - 10.1145/3427773.3427862

M3 - Article in proceedings

SN - 9781450381932

BT - RLEM'20: Proceedings of the 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities

PB - Association for Computing Machinery

T2 - 1st International Workshop on Reinforcement Learning for Energy Management in Buildings & Cities

Y2 - 17 November 2020 through 17 November 2020

ER -

Demand Response through Price-setting Multi-agent Reinforcement Learning

Abstract

Workshop

UN SDGs

Access to Document

OpenUrl availability

Fingerprint

Cite this