ARLEM : Adaptive Reinforcement Learning for Energy Management
Nik, Vahid M. LU
(2025)
19th IBPSA Conference on Building Simulation, BS 2025
In Building Simulation Conference Proceedings
19.
- Abstract
Managing energy in urban areas is complex due to the diversity of energy users, dynamic environmental conditions, and the increased instability caused by extreme weather events. This work explains the impacts of a new technology based on implanting adaptive reinforcement learning (ARL) into energy management (EM), which is called ARLEM. The technology uses an online, value-based, model-free ARL engine that periodically updates its policy by replacing less effective actions with those better suited to changing environmental conditions. Multiple policy update mechanisms are examined, differing in update frequency, duration, and action selection criteria. ARLEM controls the energy performance of a typical urban block in Madrid during a... (More)
Managing energy in urban areas is complex due to the diversity of energy users, dynamic environmental conditions, and the increased instability caused by extreme weather events. This work explains the impacts of a new technology based on implanting adaptive reinforcement learning (ARL) into energy management (EM), which is called ARLEM. The technology uses an online, value-based, model-free ARL engine that periodically updates its policy by replacing less effective actions with those better suited to changing environmental conditions. Multiple policy update mechanisms are examined, differing in update frequency, duration, and action selection criteria. ARLEM controls the energy performance of a typical urban block in Madrid during a summer in 2040-2069 with two heatwaves, considering 17 future climate scenarios. The block consists of 24 buildings with different types and ages, representing the present building stock. Results show that ARLEM enhances climate resilience by increasing energy flexibility within the network and reducing both average and peak energy demands, while affecting indoor comfort marginally. Since ARLEM does not require prior knowledge of system dynamics, it effectively copes with the complexity of buildings and urban systems, making the technology a scalable and cost-effective EM solution.
(Less)
- author
- Nik, Vahid M.
LU
- organization
- publishing date
- 2025
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- BS 2025 - Proceedings of Building Simulation 2025 : 19th Conference of IBPSA - 19th Conference of IBPSA
- series title
- Building Simulation Conference Proceedings
- volume
- 19
- publisher
- International Building Performance Simulation Association (IBPSA)
- conference name
- 19th IBPSA Conference on Building Simulation, BS 2025
- conference location
- Brisbane, Australia
- conference dates
- 2025-08-24 - 2025-08-27
- external identifiers
-
- scopus:105035251663
- ISSN
- 2522-2708
- ISBN
- 9781775052043
- DOI
- 10.26868/25222708.2025.1247
- language
- English
- LU publication?
- yes
- additional info
- Publisher Copyright: © 2025 Building Simulation Conference Proceedings. All rights reserved.
- id
- 4f169108-c5e8-4529-b7ad-6e051e95c425
- date added to LUP
- 2026-05-05 16:20:11
- date last changed
- 2026-05-12 12:40:40
@inproceedings{4f169108-c5e8-4529-b7ad-6e051e95c425,
abstract = {{<p>Managing energy in urban areas is complex due to the diversity of energy users, dynamic environmental conditions, and the increased instability caused by extreme weather events. This work explains the impacts of a new technology based on implanting adaptive reinforcement learning (ARL) into energy management (EM), which is called ARLEM. The technology uses an online, value-based, model-free ARL engine that periodically updates its policy by replacing less effective actions with those better suited to changing environmental conditions. Multiple policy update mechanisms are examined, differing in update frequency, duration, and action selection criteria. ARLEM controls the energy performance of a typical urban block in Madrid during a summer in 2040-2069 with two heatwaves, considering 17 future climate scenarios. The block consists of 24 buildings with different types and ages, representing the present building stock. Results show that ARLEM enhances climate resilience by increasing energy flexibility within the network and reducing both average and peak energy demands, while affecting indoor comfort marginally. Since ARLEM does not require prior knowledge of system dynamics, it effectively copes with the complexity of buildings and urban systems, making the technology a scalable and cost-effective EM solution.</p>}},
author = {{Nik, Vahid M.}},
booktitle = {{BS 2025 - Proceedings of Building Simulation 2025 : 19th Conference of IBPSA}},
isbn = {{9781775052043}},
issn = {{2522-2708}},
language = {{eng}},
publisher = {{International Building Performance Simulation Association (IBPSA)}},
series = {{Building Simulation Conference Proceedings}},
title = {{ARLEM : Adaptive Reinforcement Learning for Energy Management}},
url = {{http://dx.doi.org/10.26868/25222708.2025.1247}},
doi = {{10.26868/25222708.2025.1247}},
volume = {{19}},
year = {{2025}},
}