基于强化学习的地下空间除湿机组系统优化控制研究  

Optimal Control of Underground Space Dehumidifier SystemBased on Reinforcement Learning

作　　者：赵安军[1] 魏渊 张洺瑞 任启航 ZHAO Anjun;WEI Yuan;ZHANG Mingrui;REN Qihang(School of Building Services Science and Engineering,Xi’an University of Architecture and Technology,Xi’an 710000,China)

机构地区：[1]西安建筑科技大学建筑设备科学与工程学院,西安710000

出　　处：《建筑节能(中英文)》2025年第4期89-98,共10页Building Energy Efficiency

基　　金：陕西省重点研发计划(智慧型公共医疗建筑节能降碳关键技术研究与应用示范)资助项目(2023-ZDLSF-30)。

摘　　要：针对地下空间建筑在夏季常面临闷热潮湿的问题,传统的除湿设备如空调和除湿机组存在非线性和滞后的运行问题,导致能耗较高。因此,提出了一种基于强化学习的地下空间建筑除湿机组系统的节能优化控制方法。利用神经网络建立了除湿机组系统的环境模型,并将室内湿度和系统能效设定为控制目标。针对地下空间建筑除湿机组系统,构建了基于双延迟深度确定性策略梯度(Twin Delayed Deep Deterministic Policy Gradient, TD3)算法的强化学习智能体结构。通过智能体与环境模型的交互,不断尝试调整除湿机组中的蒸发温度和冷却塔风机频率,使室内湿度更接近设定的湿度值,并在一定程度上提高系统能效,从而降低能源消耗,实现一定的节能效果。In response to the issue of hot and humid in underground building spaces in summer,traditional dehumidification devices such as air conditioning and dehumidification units exhibit non-linear and lagging operational problems,leading to high energy consumption.Therefore,an energy-saving optimal control method based on reinforcement learning was proposed for the dehumidification unit system of underground space buildings.An environmental model of the dehumidifier system was established by using neural network,and the indoor humidity and system efficiency were set as control objectives.Aiming at the dehumidification unit system of underground space building,a reinforcement learning agent structure based on the TD3(Twin Delayed Deep Deterministic Policy Gradient)algorithm was constructed.Through the interaction between the agent and the environmental model,the evaporation temperature and the cooling tower fan frequency in the dehumidifier unit were continuously adjusted,which made the indoor humidity closer to the set humidity value This approach effectively enhanced system efficiency to some extent,reducing energy consumption and achieving certain energy-saving benefits.

关 键 词：地下空间建筑 除湿机组 系统能效 双延迟深度确定性策略梯度 强化学习 

分 类 号：TU83[建筑科学—供热、供燃气、通风及空调工程] TU965