基于RL/DRL改进算法的BEMS应用研究  

Research on the application of BEMS based on improved RL/DRL algorithm

作　　者：马斌 高海洋 郑馨怡 端凌立 张若微 MA Bin;GAO Haiyang;ZHENG Xinyi;DUAN Lingli;ZHANG Ruowei(Nanjing Electric Power Design and Research Institute Co.,Ltd.,Nanjing 210003,China)

机构地区：[1]南京电力设计研究院有限公司,江苏南京210003

出　　处：《粘接》2024年第11期135-138,共4页Adhesion

摘　　要：为了平衡建筑能耗和舒适度,对深度强化学习在建筑能源管理系统中的应用研究进行了回顾。阐述了建筑能源管理系统的重要性和目标,以及该研究领域传统控制方法存在的不足。介绍了深度强化学习和建筑能源管理的理论基础以及二者之间的关系和结合的可能性。本文重点回顾了深度强化学习在建筑能源管理系统(BEMS)不同方面的应用研究。根据现有相关研究中的局限性,提出了深度强化学习在建筑能源管理系统中所面临的挑战和潜在的研究方向。本文旨在通过对现有研究的回顾,为未来深度强化学习在建筑能源管理系统中的进一步应用提供一些洞见。In order to balance the energy consumption and comfort of buildings,the application of deep reinforcement learning in building energy management systems was reviewed.This paper expounds the importance and objectives of building energy management,as well as the limitations of traditional control methods in this research field.The theoretical basis of deep reinforcement learning and building energy management,as well as the relationship and possibility of combining the two,are introduced.This paper focuses on the application of deep reinforcement learning in different aspects of building energy management systems(BEMS).According to the limitations of existing related researches,the challenges and potential research directions of deep reinforcement learning in building energy management systems are proposed.The purpose of this paper is to provide some insights for the further application of deep reinforcement learning in building energy management systems in the future through a review of existing research.

关 键 词：强化学习 深度强化学习 建筑能源管理系统 控制策略 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论] TU18[自动化与计算机技术—计算机科学与技术]