面向需求响应的空调智能体:概念、构建和应用  

Air Conditioner Intelligent Agent for Demand Response:Concept,Construction,and Application

作　　者：刘伟峰 范帅 田济源 吴清 杨宜龙 何光宇[1] LIU Weifeng;FAN Shuai;TIAN Jiyuan;WU Qing;YANG Yilong;HE Guangyu(Key Laboratory of Control of Power Transmission and Conversion,Ministry of Education(Shanghai Jiao Tong University),Shanghai 200240,China;Tropical New-type Power System Research Institute,North China Electric Power University,Haikou 570311,China)

机构地区：[1]电力传输与功率变换教育部重点实验室(上海交通大学),上海市200240 [2]华北电力大学海南热带新型电力系统研究院,海南省海口市570311

出　　处：《电力系统自动化》2024年第22期70-83,共14页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目(52207123)。

摘　　要：园区内各变频空调(IAC)的运行特性、运行环境、用户行为和用户习惯千差万别,这给园区运营商引导大规模IAC参与需求响应(DR)带来了挑战。首先,对IAC的信息物理社会属性进行了分析,并提出了基于规则的社会属性刻画方法。其次,根据具身通用人工智能的概念,提出了面向DR的空调智能体概念。然后,从主动感知、主动交互、主动行动和主动演化的维度设计了空调智能体,其能在DR期间融合IAC的信息物理社会属性,使自身的负荷形状曲线接近被下发的目标形状曲线。进而,参考基于负荷准线的DR机制,提出了大规模空调智能体自下而上参与DR的方法,以实现聚沙成塔效应。最后,利用离散事件仿真框架SimPy对1万个空调智能体参与DR的情形进行仿真,从刻画社会属性的规则自演化、空调智能体性能和聚沙成塔效应这3个方面进行分析。结果表明,空调智能体在时效性、友好性、用户满意性和鲁棒性方面具有较大优势。In the park,the operation characteristics,operation environments,user behaviors and user habits of each inverter air conditioner(IAC)vary significantly,posing challenges for park operators in facilitating large-scale IAC participation in the demand response(DR).Firstly,the cyber-physical-social attributes of IAC are analyzed,and a rule-based method for characterizing social attributes is proposed.Secondly,according to the concept of embodied general artificial intelligence,the concept of inverter air conditioner intelligent agent(IAC-IA)for DR is introduced.Then,the IAC-IA is designed from the dimensions of proactive perception,proactive interaction,proactive action,and proactive evolution,enabling it to integrate the cyber-physical-social attributes of IAC during DR and align its load shape curve with the target shape curve being issued.Furthermore,referring to the DR mechanism based on customer directrix load,a bottom-up approach for the large-scale participation of IAC-IAs in DR is proposed to achieve the cumulative effect.Finally,by using the discrete event simulation framework SimPy,a simulation of 10000 IAC-IAs participating in DR is conducted for analyzing the self-evolution on rules characterizing social attributes,the performance of IAC-IA,and the cumulative effect.The results demonstrate significant advantages of IAC-IA in timeliness,friendliness,user satisfaction,and robustness.

关 键 词：空调智能体 需求响应 信息物理社会属性 负荷准线 自演化 人工智能 

分 类 号：TM73[电气工程—电力系统及自动化]