Pub Date : 2022-09-02DOI: 10.1080/19401493.2022.2104374
Weiye Huo, Mengmeng Zhao, G. Wang
The extra heat loss of a horizontal gap and a vertical gap formed by thermal insulation plate splicing is numerically studied by computational fluid dynamics. Three combinations of heat transfer modes are designed for three heat transfer modes, and the temperature, velocity distributions, and extra heat loss of the air layer in the gap under different combinations of heat transfer modes are obtained when the temperature difference is 22 K and the thickness of thermal insulation plate 300 mm, to determine the proportion of the heat loss of each heat transfer mode in the total heat loss. The results show that the main mode of heat loss in the horizontal gap is heat radiation, while the main mode of heat loss in the vertical gap is heat convection. The horizontal gap has little effect on heat loss, and the vertical gap has a great effect on heat loss.
{"title":"Analysis of gap heat loss in external thermal insulation system of passive low-energy-consumption building based on CFD","authors":"Weiye Huo, Mengmeng Zhao, G. Wang","doi":"10.1080/19401493.2022.2104374","DOIUrl":"https://doi.org/10.1080/19401493.2022.2104374","url":null,"abstract":"The extra heat loss of a horizontal gap and a vertical gap formed by thermal insulation plate splicing is numerically studied by computational fluid dynamics. Three combinations of heat transfer modes are designed for three heat transfer modes, and the temperature, velocity distributions, and extra heat loss of the air layer in the gap under different combinations of heat transfer modes are obtained when the temperature difference is 22 K and the thickness of thermal insulation plate 300 mm, to determine the proportion of the heat loss of each heat transfer mode in the total heat loss. The results show that the main mode of heat loss in the horizontal gap is heat radiation, while the main mode of heat loss in the vertical gap is heat convection. The horizontal gap has little effect on heat loss, and the vertical gap has a great effect on heat loss.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89329854","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-02DOI: 10.1080/19401493.2022.2107071
Hao Zhang, Jie Cai, J. Braun
This paper presents a holistic building life-cycle assessment methodology that estimates the embodied and operational global warming potentials (GWPs) of a building covering the envelope, mechanical and lighting systems. The methodology relies on EnergyPlus to generate the use-phase energy consumption for any given building and incorporates a streamlined procedure to extract construction materials, which are used for building envelope GWP analysis. Embodied GWP accounting was performed for a representative packaged electric cooling and gas heating system and three types of lighting technologies, i.e., incandescent, compact fluorescent (CFL) and light-emitting diode (LED). The methodology was applied for carbon footprint analysis of five U.S. Department of Energy commercial building prototypes across seven climate locations. The results show that the operation phase has a dominant contribution (more than 74%) on the overall building environmental impact. LED and CFL lighting result in 45% whole-building energy consumption and 35% GWP reductions compared to incandescent lights.
{"title":"A whole building life-cycle assessment methodology and its application for carbon footprint analysis of U.S. commercial buildings","authors":"Hao Zhang, Jie Cai, J. Braun","doi":"10.1080/19401493.2022.2107071","DOIUrl":"https://doi.org/10.1080/19401493.2022.2107071","url":null,"abstract":"This paper presents a holistic building life-cycle assessment methodology that estimates the embodied and operational global warming potentials (GWPs) of a building covering the envelope, mechanical and lighting systems. The methodology relies on EnergyPlus to generate the use-phase energy consumption for any given building and incorporates a streamlined procedure to extract construction materials, which are used for building envelope GWP analysis. Embodied GWP accounting was performed for a representative packaged electric cooling and gas heating system and three types of lighting technologies, i.e., incandescent, compact fluorescent (CFL) and light-emitting diode (LED). The methodology was applied for carbon footprint analysis of five U.S. Department of Energy commercial building prototypes across seven climate locations. The results show that the operation phase has a dominant contribution (more than 74%) on the overall building environmental impact. LED and CFL lighting result in 45% whole-building energy consumption and 35% GWP reductions compared to incandescent lights.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78521813","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2099465
Suroor M. Dawood, A. Hatami, R. Homod
This paper presents Model-based Reinforcement Learning (MB-RL) techniques to control the indoor air temperature, and CO2 concentration level, and minimize the energy consumption of the heating, ventilating, and air conditioning (HVAC) systems, simultaneously. For this purpose, a trade-off is made between maintaining indoor comfort levels and minimizing energy consumption. The control of the HVAC system is performed using the Deterministic Policy RL (DP-RL) method. Moreover, the nonlinear autoregressive exogenous neural network (NARX-NN) is employed as an approximation function with DP-RL method to provide a hybrid DP-NARX-RL controller. By applying the DP-RL and DP-NARX-RL controllers to the HVAC system of a typical building, parameters such as the indoor comfort levels, the electrical power, and energy consumed, and the energy costs at various pricing schemes are evaluated for two case studies. In both cases, the results show the better performance of DP-NARX-RL compared to DP-RL, RL, and PID controllers.
{"title":"Trade-off decisions in a novel deep reinforcement learning for energy savings in HVAC systems","authors":"Suroor M. Dawood, A. Hatami, R. Homod","doi":"10.1080/19401493.2022.2099465","DOIUrl":"https://doi.org/10.1080/19401493.2022.2099465","url":null,"abstract":"This paper presents Model-based Reinforcement Learning (MB-RL) techniques to control the indoor air temperature, and CO2 concentration level, and minimize the energy consumption of the heating, ventilating, and air conditioning (HVAC) systems, simultaneously. For this purpose, a trade-off is made between maintaining indoor comfort levels and minimizing energy consumption. The control of the HVAC system is performed using the Deterministic Policy RL (DP-RL) method. Moreover, the nonlinear autoregressive exogenous neural network (NARX-NN) is employed as an approximation function with DP-RL method to provide a hybrid DP-NARX-RL controller. By applying the DP-RL and DP-NARX-RL controllers to the HVAC system of a typical building, parameters such as the indoor comfort levels, the electrical power, and energy consumed, and the energy costs at various pricing schemes are evaluated for two case studies. In both cases, the results show the better performance of DP-NARX-RL compared to DP-RL, RL, and PID controllers.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83092259","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2102676
Xavier Marsault
In generative design, it is imperative for an architect to evaluate very quickly the performance of many buildings produced. Knowing in interactive time the daylighting potential of a generated form at an early stage of its design, with a minimum of parameters, allows to quickly choose among many variants. The daylight factor computational metamodel presented here in the case of modular buildings allows to instantly compare these solutions in order to make judicious choices in dimensioning, without performing time-consuming simulations. Another challenge was to achieve realtime computation for the daylight factor without using a GPU. We have addressed this objective via an hybrid computation both based on physical and statistical modeling, and on a physical-based computation engine specifically used for the optimization of buildings composed of multiple living units. We detail the full implementation in a generative design software leading to impressive computation times of the order of one ms.
{"title":"Achieving realtime daylight factor computation for modular buildings in generative design","authors":"Xavier Marsault","doi":"10.1080/19401493.2022.2102676","DOIUrl":"https://doi.org/10.1080/19401493.2022.2102676","url":null,"abstract":"In generative design, it is imperative for an architect to evaluate very quickly the performance of many buildings produced. Knowing in interactive time the daylighting potential of a generated form at an early stage of its design, with a minimum of parameters, allows to quickly choose among many variants. The daylight factor computational metamodel presented here in the case of modular buildings allows to instantly compare these solutions in order to make judicious choices in dimensioning, without performing time-consuming simulations. Another challenge was to achieve realtime computation for the daylight factor without using a GPU. We have addressed this objective via an hybrid computation both based on physical and statistical modeling, and on a physical-based computation engine specifically used for the optimization of buildings composed of multiple living units. We detail the full implementation in a generative design software leading to impressive computation times of the order of one ms.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91315966","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2075939
J. Baek, Hansaem Park, Seongju Chang
Unlike in the past, as the scale of buildings expands from a single building to a group of buildings, it is necessary to respond to the demand for energy consumption for the entire community. This study attempted to improve the electricity consumption prediction performance of single buildings by reflecting the inherent occupant behaviour patterns embedded in the shared electricity consumption data of single buildings. The proposed method utilized electricity consumption data for all other 7 target buildings as a part of LSTM model inputs in combination with meteorological data to predict the electricity consumption profile of a specific building. As a result, the proposed method reduced the error by utilizing the similar energy profiles of the surrounding buildings, *and the prediction performance of the total electricity consumption of the community was increased by 5.11% based on RMSE. The newly proposed method reflects the inherent occupant behaviour patterns and temporal electricity consumption patterns from energy profiles of nearby buildings.
{"title":"Enhanced LSTM-based community energy consumption prediction model leveraging shared building cluster datasets","authors":"J. Baek, Hansaem Park, Seongju Chang","doi":"10.1080/19401493.2022.2075939","DOIUrl":"https://doi.org/10.1080/19401493.2022.2075939","url":null,"abstract":"Unlike in the past, as the scale of buildings expands from a single building to a group of buildings, it is necessary to respond to the demand for energy consumption for the entire community. This study attempted to improve the electricity consumption prediction performance of single buildings by reflecting the inherent occupant behaviour patterns embedded in the shared electricity consumption data of single buildings. The proposed method utilized electricity consumption data for all other 7 target buildings as a part of LSTM model inputs in combination with meteorological data to predict the electricity consumption profile of a specific building. As a result, the proposed method reduced the error by utilizing the similar energy profiles of the surrounding buildings, *and the prediction performance of the total electricity consumption of the community was increased by 5.11% based on RMSE. The newly proposed method reflects the inherent occupant behaviour patterns and temporal electricity consumption patterns from energy profiles of nearby buildings.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84633092","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2082531
Adam Neale, M. Kummert, M. Bernier
A new bottom-up white-box building stock energy modelling approach is presented in this paper. The model is a physics simulation-based stock model that can be used to compare the base case building stock with technological variations for comparative assessments. The model accuracy is compared to known stock data for a variety of categories, including end-use, energy source and building type. The model predicts the energy use of a building stock in a Canadian region with good agreement across all categories, with the total energy consumption of the model within 1.5% of the real stock energy use. Our analysis shows that modelling a sufficiently large sample of stock is required to reduce the expected deviation for lesser-represented portions of the stock. A case study illustrates how the model can be applied to compare different heating system distributions, in order to assess their impact on greenhouse gas emissions and peak electricity demand.
{"title":"Development of a bottom-up white-box building stock energy model for single-family dwellings","authors":"Adam Neale, M. Kummert, M. Bernier","doi":"10.1080/19401493.2022.2082531","DOIUrl":"https://doi.org/10.1080/19401493.2022.2082531","url":null,"abstract":"A new bottom-up white-box building stock energy modelling approach is presented in this paper. The model is a physics simulation-based stock model that can be used to compare the base case building stock with technological variations for comparative assessments. The model accuracy is compared to known stock data for a variety of categories, including end-use, energy source and building type. The model predicts the energy use of a building stock in a Canadian region with good agreement across all categories, with the total energy consumption of the model within 1.5% of the real stock energy use. Our analysis shows that modelling a sufficiently large sample of stock is required to reduce the expected deviation for lesser-represented portions of the stock. A case study illustrates how the model can be applied to compare different heating system distributions, in order to assess their impact on greenhouse gas emissions and peak electricity demand.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80403818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2094466
Jerson Sanchez, Zhiming Jiang, Jie Cai
This paper presents a modelling methodology to characterize heating, ventilation and air-conditioning (HVAC) equipment lifetime impact of load controls and an aging-aware demand responsive control strategy for single-stage HVAC systems to strike a balance between the electric utility and HVAC life-cycle costs. To assess the control performance and evaluate potential trade-offs between energy consumption, utility cost and equipment lifetime impact, whole-month simulation tests for a single-zone office building have been conducted for the proposed aging-aware control strategy along with two benchmarking strategies – energy minimizing and utility cost minimizing controllers. Test results show that the aging-aware demand response strategy could result in reductions of HVAC equipment aging effect by 18.8% to 39.1% compared to the utility-priority controller. The total building operation cost, with the electricity utility and HVAC life-cycle costs combined, could be reduced by up to 13.2% compared to the utility minimizing strategy and by up to 16.1% compared to the energy minimization baseline.
{"title":"Modelling and mitigating lifetime impact of building demand responsive control of heating, ventilation and air-conditioning systems","authors":"Jerson Sanchez, Zhiming Jiang, Jie Cai","doi":"10.1080/19401493.2022.2094466","DOIUrl":"https://doi.org/10.1080/19401493.2022.2094466","url":null,"abstract":"This paper presents a modelling methodology to characterize heating, ventilation and air-conditioning (HVAC) equipment lifetime impact of load controls and an aging-aware demand responsive control strategy for single-stage HVAC systems to strike a balance between the electric utility and HVAC life-cycle costs. To assess the control performance and evaluate potential trade-offs between energy consumption, utility cost and equipment lifetime impact, whole-month simulation tests for a single-zone office building have been conducted for the proposed aging-aware control strategy along with two benchmarking strategies – energy minimizing and utility cost minimizing controllers. Test results show that the aging-aware demand response strategy could result in reductions of HVAC equipment aging effect by 18.8% to 39.1% compared to the utility-priority controller. The total building operation cost, with the electricity utility and HVAC life-cycle costs combined, could be reduced by up to 13.2% compared to the utility minimizing strategy and by up to 16.1% compared to the energy minimization baseline.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75414199","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2092652
S. Anbarasu, W. Zuo, Yangyang Fu, Yash Shukla, Rajan Rawal
Direct evaporative coolers (DECs) are a low-energy cooling alternative to conventional air conditioning in hot-dry climates. The key component of DEC is the cooling pad, which evaporatively cools the air passing through it. While detailed numerical models of heat and mass transfer have been proposed for the cooling pad, these require many input parameters that are not readily accessible. Alternatively, simplified models lack accuracy and are confined to common types of cooling pad. To address these limitations, we developed and validated a physics-based model, that only needs the nominal data to compute the heat and mass transfer with considerable accuracy. The proposed model is implemented in Modelica, an equation-based object-oriented modeling language. For comparison, a basic lumped model from EnergyPlus based on the efficiency curve of the cooling pad is also implemented. The physics-based model exhibits <2% error from the experimental data and the lumped model exhibits a 12.3% error.
{"title":"Validated open-source Modelica model of direct evaporative cooler with minimal inputs","authors":"S. Anbarasu, W. Zuo, Yangyang Fu, Yash Shukla, Rajan Rawal","doi":"10.1080/19401493.2022.2092652","DOIUrl":"https://doi.org/10.1080/19401493.2022.2092652","url":null,"abstract":"Direct evaporative coolers (DECs) are a low-energy cooling alternative to conventional air conditioning in hot-dry climates. The key component of DEC is the cooling pad, which evaporatively cools the air passing through it. While detailed numerical models of heat and mass transfer have been proposed for the cooling pad, these require many input parameters that are not readily accessible. Alternatively, simplified models lack accuracy and are confined to common types of cooling pad. To address these limitations, we developed and validated a physics-based model, that only needs the nominal data to compute the heat and mass transfer with considerable accuracy. The proposed model is implemented in Modelica, an equation-based object-oriented modeling language. For comparison, a basic lumped model from EnergyPlus based on the efficiency curve of the cooling pad is also implemented. The physics-based model exhibits <2% error from the experimental data and the lumped model exhibits a 12.3% error.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91287604","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2095032
M. Ismaiel, L. Westover, Yuxiang Chen
The effective thermal resistance of masonry cavity walls is affected by thermal bridging, which is common in connecting masonry veneers to structural backup walls. Therefore, a precise estimation of the R-value of masonry cavity walls is currently a time-consuming task, which lengthens the design process, especially in the early design stage where many design options (e.g. structural and thermal) need to be explored holistically. This paper presents an efficient approach for estimating the R-values of common masonry cavity wall configurations in the form of simple design charts and R-value multipliers. Parameters such as the block density, thermal insulation value, and the types of ties and shelf angles are addressed in this study. The approach simultaneously provides the mechanical (the masonry compressive strength, fm′) and thermal (R-value) properties of different cavity wall configurations, allowing designers to obtain appropriate structural and thermal properties during a preliminary design phase without using computer simulations.
{"title":"An efficient approach for thermal design of masonry walls using design charts and R-value multipliers","authors":"M. Ismaiel, L. Westover, Yuxiang Chen","doi":"10.1080/19401493.2022.2095032","DOIUrl":"https://doi.org/10.1080/19401493.2022.2095032","url":null,"abstract":"The effective thermal resistance of masonry cavity walls is affected by thermal bridging, which is common in connecting masonry veneers to structural backup walls. Therefore, a precise estimation of the R-value of masonry cavity walls is currently a time-consuming task, which lengthens the design process, especially in the early design stage where many design options (e.g. structural and thermal) need to be explored holistically. This paper presents an efficient approach for estimating the R-values of common masonry cavity wall configurations in the form of simple design charts and R-value multipliers. Parameters such as the block density, thermal insulation value, and the types of ties and shelf angles are addressed in this study. The approach simultaneously provides the mechanical (the masonry compressive strength, fm′) and thermal (R-value) properties of different cavity wall configurations, allowing designers to obtain appropriate structural and thermal properties during a preliminary design phase without using computer simulations.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79473036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-04DOI: 10.1080/19401493.2022.2103586
S. Cho, C. Park
Although it is widely acknowledged that reinforcement learning (RL) can be beneficial for building control, many RL-based control actions remain unexplainable in the daily practice of facility managers. This paper reports a rule reduction framework using explainable RL to enhance the practicality of the control strategy. First, deep Q-learning was applied to explore the optimal control strategies of a parallel cooling system (ice-based thermal system + geothermal heat pump system) of an existing office building. A set of modularized and interconnected data-driven models was developed using ANNs for pretraining an artificial agent. After exploring the control strategies, the decision-making rules of the agent were reduced using a decision tree. The performance of the reduced-order rule-based control proved comparable to the complex and uninterpretable control strategy of deep Q-learning. The difference in energy savings between the two is marginal at 1.2%.
{"title":"Rule reduction for control of a building cooling system using explainable AI","authors":"S. Cho, C. Park","doi":"10.1080/19401493.2022.2103586","DOIUrl":"https://doi.org/10.1080/19401493.2022.2103586","url":null,"abstract":"Although it is widely acknowledged that reinforcement learning (RL) can be beneficial for building control, many RL-based control actions remain unexplainable in the daily practice of facility managers. This paper reports a rule reduction framework using explainable RL to enhance the practicality of the control strategy. First, deep Q-learning was applied to explore the optimal control strategies of a parallel cooling system (ice-based thermal system + geothermal heat pump system) of an existing office building. A set of modularized and interconnected data-driven models was developed using ANNs for pretraining an artificial agent. After exploring the control strategies, the decision-making rules of the agent were reduced using a decision tree. The performance of the reduced-order rule-based control proved comparable to the complex and uninterpretable control strategy of deep Q-learning. The difference in energy savings between the two is marginal at 1.2%.","PeriodicalId":49168,"journal":{"name":"Journal of Building Performance Simulation","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2022-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90058978","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：