基于强化学习的HVAC能耗最小化控制器设计OA
Reinforcement Learning Based Controller Design for HVAC Energy Minimization
针对暖通空调(HVAC)系统能耗优化这个问题,提出一种基于深度强化学习的能耗最小化控制方法.通过建立HVAC系统动态数学模型的方式,设计深度Q网络(DQN)控制算法,构建包含温度与湿度及CO2浓度等多维状态空间,制定融合用户舒适度与能源效率的奖励函数机制.试验结果表明,相较传统控制方案,该控制器在系统稳定性与调节性能方面具备显著优势,为大型建筑暖通空调系统节能减排开辟创新途径.
Aiming at the problem of optimizing the energy consumption of heating,ventilation and air conditioning(HVAC)systems,an energy minimization control method based on deep reinforcement learning is proposed.By establishing a dynamic mathematical model of the HVAC system,designing a deep Q-network(DQN)control algorithm,constructing a multi-dimensional state space containing temperature,humidity and CO2 concentration,and formulating a reward function mechanism that integrates user comfort and energy efficiency.The experimental results show that compared with the traditional control scheme,the controller has significant advantages in system stability and regulation performance,which opens up an innovative way for energy saving and emission reduction in HVAC systems of large buildings.
李洋;徐航;程彬;陈可
南阳理工学院,河南南阳 473000南阳理工学院,河南南阳 473000南阳理工学院,河南南阳 473000南阳理工学院,河南南阳 473000
土木建筑
强化学习HVAC系统数学建模DQN算法能耗优化
reinforcement learningHVAC systemmathematical modelingDQN algorithmenergy consumption optimization
《数码设计》 2025 (8)
115-118,4
1.程彬,项目名称:适应多种实验教学管理模式的数字化平台建设研究(项目编号:24KJGG095)2.陈可,河南省科技攻关项目,项目名称:基于隐私计算的强直性脊柱炎全基因组数据关联分析与共享研究(项目编号:232102211058).
评论