摘要
室内热环境在影响居住者的舒适度和生产力方面起着至关重要的作用.供暖、通风和空调(HVAC)系统是住宅建筑中控制室内热环境的主要手段.然而,HVAC系统消耗大量能源,占建筑物总能源消耗的20%~40%,导致高昂的运营成本.因此,设计能够在保持最佳热环境并降低能源消耗的HVAC控制策略至关重要.对此,提出一种基于增强智能(AuI)的框架,结合人类指导和深度强化学习(DRL)实现HVAC控制优化.首先,建立优化问题,旨在平衡能源消耗和热舒适度,并最小化两者成本;其次,构建人类行为模型,模拟不同热环境下居住者的反应;然后,提出AuI增强深度确定性策略梯度(DDPG)算法,用于学习适配热舒适度和能源效率的最佳HVAC策略;最后,搭建了仿真实验,用于策略训练和性能评估.实验结果表明,相较于无人类指导的对比算法,AuI-DDPG能减少38.7%的HVAC能耗,同时改善31.1%的居住者热舒适.
The indoor thermal environment plays a crucial role in influencing occupants'comfort and productivity.Heating,ventilation,and air conditioning(HVAC)systems are the primary means of controlling indoor thermal conditions in residential buildings.However,HVAC systems consume a significant amount of energy,accounting for 20%–40%of a building's total energy usage,resulting in high operational costs.Therefore,it is essential to design HVAC control strategies that minimize energy consumption while maintaining an optimal thermal environment.This paper proposes an augmented intelligence(AuI)-based framework that integrates human guidance with deep reinforcement learning(DRL)to optimize HVAC control.We formulate an optimization problem aimed at balancing energy consumption with thermal comfort,minimizing the costs associated with both factors.A human behavior model is developed to simulate occupant responses to varying thermal conditions.The proposed AuI-enhanced deep deterministic policy gradient(DDPG)algorithm is employed to learn the optimal HVAC strategy,considering both thermal comfort and energy efficiency.A simulation environment is constructed for strategy training and performance evaluation.The results show that,compared to a baseline without human guidance,the AuI-DDPG algorithm reduces HVAC energy consumption by 38.7%while improving occupant thermal comfort by 31.1%.
作者
魏传锋
洪春森
何鹏
崔亚平
吴大鹏
王汝言
周圣博
WEI Chuan-feng;HONG Chun-sen;HE Peng;CUI Ya-ping;WU Da-peng;WANG Ru-yan;ZHOU Sheng-bo(School of Communications and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China;Advanced Network and Intelligent Connection Technology Key Laboratory of Chongqing Education Commission of China,Chongqing 400065,China;Chongqing Key Laboratory of Ubiquitous Sensing and Networking,Chongqing 400065,China;China Satellite Network Network System Research Institute,Beijing 100094,China)
出处
《控制与决策》
北大核心
2025年第12期3551-3564,共14页
Control and Decision
基金
国家自然科学基金项目(61801065,62271096,61871062,U20A20157,62061007,U24A20211)
重庆市教委科学技术研究项目(KJQN202000603,KJQN202300621)
重庆市自然科学基金项目(CSTB2024NSCQLZX0124,CSTB2022NSCQ-MSX0468,cstc2020jcyjzdxmX0024,CSTB2023NSCQ-LZX0134)
重庆市高校创新研究群体资助项目(CXQT20017)
重邮信通青创团队支持计划项目(SCIE-QN-2022-04)
四川省重点研发计划项目(2024YFHZ0093)。
关键词
暖通空调
热舒适控制
增强人工智能
深度强化学习
heating,ventilation and air conditioning
thermal comfort control
augmented intelligence
deep reinforcement learning
