摘要
空调系统运行优化是建筑节能的重要组成部分。提出了在空调系统运行优化中应用强化学习算法,主要采用拟合Q迭代算法。结合空调系统运行优化的实际需求及强化学习算法的特征,将强化学习控制器在空调系统运行优化中的应用过程分为4个阶段,包括准备、初始、探索和运行阶段,并具体描述了各阶段完整的算法流程。通过强化学习控制器在空气源热泵联合电辅热系统中的应用,案例对其控制效果进行验证分析。仿真结果显示,提出的基于强化学习算法的空调系统运行方法在满足建筑负荷需求的同时,可以有效降低建筑运行费用。与模型预测控制方法相比,强化学习控制器响应速度更快。此外,该方法具备对先验知识依赖程度低、自适应性强的特点,具备一定的实用性,其应用有助于实现空调系统精细化运行的目标。
The operation optimization of HVAC system is an essential part of building energy conservation.In this study,HVAC system operation optimization based on reinforcement learning controller is proposed.The fitted Q-iteration algorithm is selected as the main algorithm.Considering the features of the algorithm and the requirement of HVAC system operation,the application of RLC is divided into four phases,including preliminary phase,initial phase,exploration phase and operation phase.Detailed process of each phase is described.The application case of RLC in air-source heat pump with auxiliary heating system for small residential building is conducted to validate the performance of the reinforcement learning controller.The simulation result shows that the proposed RLC method for HVAC system operation can reduce energy cost and meet the building cooling/heating demand at the same time.Compared with model predictive control,RLC also saves the computational time for decision making.In addition,the proposed RLC method requires only little prior knowledge and is capable to adapt to the environment.The proposed method is practical and its application could enhance HVAC system operation.
作者
丁志梁
潘毅群(指导)
谢建彤
王尉同
黄治钟
DING Zhi-liang;PAN Yi-qun;XIE Jian-tong;WANG Wei-tong;HUANG Zhi-zhong(School of Mechanical Engineering,Tongji University,Shanghai 201804,China;Sino-German College of Applied Sciences,Tongji University,Shanghai 201804,China)
出处
《建筑节能》
CAS
2020年第7期14-20,共7页
BUILDING ENERGY EFFICIENCY
关键词
空调系统
运行优化
强化学习
数据驱动
HVAC system
operation optimization
reinforcement learning
data-driven
