Enhancing the Performance of Multi-Agent Reinforcement Learning for Controlling HVAC Systems

被引:11
作者
Bayer, Daniel [1 ]
Pruckner, Marco [1 ]
机构
[1] Friedrich Alexander Univ Erlangen Nurnberg, Energy Informat, Comp Sci 7, Erlangen, Germany
来源
2022 IEEE CONFERENCE ON TECHNOLOGIES FOR SUSTAINABILITY (SUSTECH) | 2022年
关键词
Multi-Agent Reinforcement Learning; Independent Q-Learning; Shared Parameters; HVAC Systems; Building Controls; Energy Efficiency; Energy Saving Controls;
D O I
10.1109/SusTech53338.2022.9794179
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Systems for heating, ventilation and air-conditioning (HVAC) of buildings are traditionally controlled by a rule-based approach. In order to reduce the energy consumption and the environmental impact of HVAC systems more advanced control methods such as reinforcement learning are promising. Reinforcement learning (RL) strategies offer a good alternative, as user feedback can be integrated more easily and presence can also be incorporated. Moreover, multi-agent RL approaches scale well and can be generalized. In this paper, we propose a multi-agent RL framework based on existing work that learns reducing on one hand energy consumption by optimizing HVAC control and on the other hand user feedback by occupants about uncomfortable room temperatures. Second, we show how to reduce training time required for proper RL-agent-training by using parameter sharing between the multiple agents and apply different pretraining techniques. Results show that our framework is capable of reducing the energy by around 6% when controlling a complete building or 8% for a single room zone. The occupants complaints are acceptable or even better compared to a rule-based baseline. Additionally, our performance analysis show that the training time can be drastically reduced by using parameter sharing.
引用
收藏
页码:187 / 194
页数:8
相关论文
共 28 条
  • [1] Reinforcement learning for whole-building HVAC control and demand response
    Azuatalam, Donald
    Lee, Wee-Lih
    de Nijs, Frits
    Liebman, Ariel
    [J]. ENERGY AND AI, 2020, 2
  • [2] Ba JL., 2016, ARXIV
  • [3] A MARKOVIAN DECISION PROCESS
    BELLMAN, R
    [J]. JOURNAL OF MATHEMATICS AND MECHANICS, 1957, 6 (05): : 679 - 684
  • [4] Beltran Alex, 2014, P 1 ACM C EMBEDDED S, P168, DOI DOI 10.1145/2674061.2674072
  • [5] Deep reinforcement learning to optimise indoor temperature control and heating energy consumption in buildings
    Brandi, Silvio
    Piscitelli, Marco Savino
    Martellacci, Marco
    Capozzoli, Alfonso
    [J]. ENERGY AND BUILDINGS, 2020, 224
  • [6] Crawley DB, 2000, ASHRAE J, V42, P49
  • [7] Federal Ministry of the Environment Nature Conservation and Nuclear Safety Berlin, 2020, CLIM PROT NUMB
  • [8] Foerster JN, 2017, PR MACH LEARN RES, V70
  • [9] GLOBALABC,, 2018, 2018 GLOB STAT REP, DOI 10.1038/s41370-017-0014-9, p. 325, 2018
  • [10] Multi-agent deep reinforcement learning: a survey
    Gronauer, Sven
    Diepold, Klaus
    [J]. ARTIFICIAL INTELLIGENCE REVIEW, 2022, 55 (02) : 895 - 943