人工智能技术在建筑能源管理中的应用场景  

Application Scenario of Artificial Intelligence Technology in Building Energy Management

作　　者：龙惟定[1] LONG Weiding(Sino-German College of Applied Sciences,Tongji University,Shanghai 201804,China)

机构地区：[1]同济大学中德工程学院,上海201804

出　　处：《建筑科学》2021年第2期127-136,145,共11页Building Science

摘　　要：本文简要介绍了建筑能源管理(BEM)的概念.并从5个方面阐述了BEM对人工智能(AI)技术的需求,即楼宇控制需要由从顶到底的基于物理模型的控制模式,转变为从底到顶的基于数据的控制模式;建筑能源系统由单一能源转变为多能源,需要解决可变可再生能源的不稳定问题;BEM的管理模型,需要从白箱转变为灰箱,甚至黑箱.此外还有负荷预测问题和非结构化数据的处理问题.文章还提出了BEM系统架构、迁移学习、物联网构建、AI与BIM的关系,以及负荷反推等需要研究的问题.文章并对人工智能在BEM领域的发展提出了建议.The paper briefly introduced the concept of Building Energy Management( BEM),and elaborated on BEM’s demands for artificial intelligence( AI) technology from five aspects,that is,building control needs to change from a top-to-bottom physical model-based control mode to a bottom-to-top data-based control mode;the transformation from a single energy source to a multi-energy source needs to solve the instability of variable renewable energy;the management model of BEM needs to change from a white box to a gray box,or even a black box,as well as load forecasting issues and unstructured data processing issues. This paper also proposed issues to be studied,such as BEM system architecture,transfer learning,construction of IoT,the relationship between AI and BIM,and backcasting of load. This paper also put forward suggestions on the development of artificial intelligence in the BEM field.

关 键 词：建筑能源管理 人工智能 负荷预测 负荷反推 物联网 多代理系统 信息物理系统 

分 类 号：TU831[建筑科学—供热、供燃气、通风及空调工程]