基于傅里叶变换的多联机空调负荷提取及预测方法  

Load extraction and load forecasting method of VRF air conditioners based on Fourier transform

作　　者：宋文杰[1] 林欣怡 张恒 田喆[1] 牛纪德[1] 夏兴祥 车闫瑾 Song Wenjie;Lin Xinyi;Zhang Heng;Tian Zhe;Niu Jide;Xia Xingxiang;Che Yanjin(Tianjin University,Tianjin;Qingdao Hisense Hitachi Air-conditioning Systems Company Limited,Qingdao)

机构地区：[1]天津大学,天津 [2]青岛海信日立空调系统有限公司,青岛

出　　处：《暖通空调》2024年第5期53-57,共5页Heating Ventilating & Air Conditioning

基　　金：国家自然科学基金项目“基于时序数据逆向解析的建筑耗热量热流组分定向提取方法”(编号:51778407)。

摘　　要：多联机空调实际制冷量通常不等同于房间真实的空调负荷需求,导致基于数据驱动的负荷预测模型的预测效果往往较差。本文提出了一种基于傅里叶变换的从实际制冷量数据中提取空调真实负荷数据的方法。案例预测结果表明,该方法能够有效提取空调负荷数据,并能将基于人工神经网络的超短期空调负荷预测模型精度提升10.94%。The actual cooling capacity of variable refrigerant flow(VRF)air conditioners is usually not equivalent to the real air conditioning load demand of room,resulting in the poor forecasting effect of data-driven load forecasting models.In this paper,a method based on Fourier transform is proposed to extract the real load of air conditioner from the actual cooling capacity.The case forecasting results show that the proposed method can effectively extract the air conditioning load data and improve the accuracy of the ultra-short-term air conditioning load forecasting models based on artificial neural networks by 10.94%.

关 键 词：多联机空调 制冷量 负荷预测 数据驱动 傅里叶变换 

分 类 号：TU831[建筑科学—供热、供燃气、通风及空调工程]