多联机空调系统在线能耗测量方法研究  

Research on online energy consumption metering methods for variable refrigerant flow air conditioning systems

作　　者：石靖峰 肖寒松 任兆亭 孙铁军 阮岱玮 石文星[3] Shi Jingfeng;Xiao Hansong;Ren Zhaoting;Sun Tiejun;Ruan Daiwei;Shi Wenxing(Tsinghua University,Beijing;Qingdao Hisense Hitachi Air-conditioning Systems Co.,Ltd.,Qingdao)

机构地区：[1]清华大学,北京100084 [2]青岛海信日立空调系统有限公司,青岛 [3]清华大学建筑技术科学系,北京100084

出　　处：《暖通空调》2023年第2期51-56,63,共7页Heating Ventilating & Air Conditioning

基　　金：青岛西海岸新区自主创新重大专项项目(编号:2020-11)。

摘　　要：实时运行能耗是实现多联机节能运行和智能控制的基础数据,但因成本原因,实际运行的多联机空调系统很少安装功率表对其能耗进行实时监测,因此研发低成本且具有工程精度的实时能耗测量方法对于多联机的节能降碳具有重要意义。本文提出了适用于单片机运行的频率-电流曲面拟合法和适用于云端大数据运算的神经网络自学习法,实验结果表明,2种在线能耗测量算法均具有良好的测量精度,在名义制冷、名义制热、最大制冷、低温制热和除霜等工况及在各种工况下自由切换连续45 h运行时,对多联机系统总功率的测量误差均在±5%以内。Real-time operation energy consumption is the basic data of the energy-saving operation and online smart control for variable refrigerant flow(VRF)systems.However,power meter is rarely installed on VRF systems for energy consumption monitoring in real projects due to their high cost.As a result,research on economy and accurate real-time energy consumption measuring method for VRF systems is of great significance for energy saving and carbon reduction.In this paper,a frequency-current surface fitting method in single chip microcomputer cases and a neural network self-learning method in cloud big data cases are proposed.Experimental results illustrate that both two methods show high measurement accuracy.The measurement errors of the total power of the VRF system distribute within±5%in various operating conditions including nominal cooling,nominal heating,maximum cooling,low temperature heating,defrosting and 45 h free switching operating conditions.

关 键 词：多联机 能耗 在线测量 数据拟合 神经网络 

分 类 号：TU831[建筑科学—供热、供燃气、通风及空调工程] TU111.195