A deep reinforcement learning approach to using whole building energy model for HVAC optimal control

Zhiang Zhang; Adrian Chong; Yuqi Pan; Chenlu Zhang; Siliang Lu; Khee Poh Lam

A deep reinforcement learning approach to using whole building energy model for HVAC optimal control

Zhiang Zhang, Adrian Chong, Yuqi Pan, Chenlu Zhang, Siliang Lu, Khee Poh Lam

Research output: Journal Publication › Conference article › peer-review

50 Citations (Scopus)

Abstract

Whole building energy model (BEM) is difficult to be used in the classical model-based optimal control (MOC) because of its high-dimension nature and intensive computational speed. This study proposes a novel deep reinforcement learning framework to use BEM for MOC of HVAC systems. A case study based on a real office building in Pennsylvania is presented in this paper to demonstrate the workflow, including building modeling, model calibration and deep reinforcement learning training. The learned optimal control policy can potentially achieve 15% of heating energy saving by simply controlling the heating system supply water temperature.

Original language	English
Pages (from-to)	675-682
Number of pages	8
Journal	ASHRAE and IBPSA-USA Building Simulation Conference
Publication status	Published - 2018
Externally published	Yes
Event	2018 ASHRAE/IBPSA-USA Building Simulation Conference: Building Performance Modeling, SimBuild 2018 - Chicago, United States Duration: 26 Sept 2018 → 28 Sept 2018

ASJC Scopus subject areas

Architecture
Modelling and Simulation

UN SDGs

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

Cite this

@article{0a70da3e4e26443388f72bc06cde86a2,

title = "A deep reinforcement learning approach to using whole building energy model for HVAC optimal control",

abstract = "Whole building energy model (BEM) is difficult to be used in the classical model-based optimal control (MOC) because of its high-dimension nature and intensive computational speed. This study proposes a novel deep reinforcement learning framework to use BEM for MOC of HVAC systems. A case study based on a real office building in Pennsylvania is presented in this paper to demonstrate the workflow, including building modeling, model calibration and deep reinforcement learning training. The learned optimal control policy can potentially achieve 15% of heating energy saving by simply controlling the heating system supply water temperature.",

author = "Zhiang Zhang and Adrian Chong and Yuqi Pan and Chenlu Zhang and Siliang Lu and Lam, {Khee Poh}",

note = "Publisher Copyright: {\textcopyright} 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us).; 2018 ASHRAE/IBPSA-USA Building Simulation Conference: Building Performance Modeling, SimBuild 2018 ; Conference date: 26-09-2018 Through 28-09-2018",

year = "2018",

language = "English",

pages = "675--682",

journal = "ASHRAE and IBPSA-USA Building Simulation Conference",

issn = "2574-6308",

publisher = "American Society of Heating Refrigerating and Air-Conditioning Engineers",

}

TY - JOUR

T1 - A deep reinforcement learning approach to using whole building energy model for HVAC optimal control

AU - Zhang, Zhiang

AU - Chong, Adrian

AU - Pan, Yuqi

AU - Zhang, Chenlu

AU - Lu, Siliang

AU - Lam, Khee Poh

PY - 2018

Y1 - 2018

N2 - Whole building energy model (BEM) is difficult to be used in the classical model-based optimal control (MOC) because of its high-dimension nature and intensive computational speed. This study proposes a novel deep reinforcement learning framework to use BEM for MOC of HVAC systems. A case study based on a real office building in Pennsylvania is presented in this paper to demonstrate the workflow, including building modeling, model calibration and deep reinforcement learning training. The learned optimal control policy can potentially achieve 15% of heating energy saving by simply controlling the heating system supply water temperature.

AB - Whole building energy model (BEM) is difficult to be used in the classical model-based optimal control (MOC) because of its high-dimension nature and intensive computational speed. This study proposes a novel deep reinforcement learning framework to use BEM for MOC of HVAC systems. A case study based on a real office building in Pennsylvania is presented in this paper to demonstrate the workflow, including building modeling, model calibration and deep reinforcement learning training. The learned optimal control policy can potentially achieve 15% of heating energy saving by simply controlling the heating system supply water temperature.

UR - http://www.scopus.com/inward/record.url?scp=85103696694&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:85103696694

SN - 2574-6308

SP - 675

EP - 682

JO - ASHRAE and IBPSA-USA Building Simulation Conference

JF - ASHRAE and IBPSA-USA Building Simulation Conference

T2 - 2018 ASHRAE/IBPSA-USA Building Simulation Conference: Building Performance Modeling, SimBuild 2018

Y2 - 26 September 2018 through 28 September 2018

ER -

A deep reinforcement learning approach to using whole building energy model for HVAC optimal control

Abstract

ASJC Scopus subject areas

UN SDGs

Other files and links

Fingerprint

Cite this