摘要: |
在变制冷剂量(VRF)空调系统中,压缩机回液将导致能量损失。本文结合大数据提出了一种基于PCA-Clustering的压缩机回液故障诊断的方法。首先提取出故障相关变量,并通过数据预处理,剔除异常值与空值;然后将处理后的数据进行主成分分析(PCA),获取降维后的新主元变量数据;最后将新的主元变量进行聚类分析(Clustering analysis)得到回液故障数据分类标签。结果表明:该方法能够在数据标签未知的情况下,较好的区分不同类别的压缩机回液故障以及正常数据,使压缩机回液故障诊断率达到94.29%。 |
关键词: 多联机系统 压缩机回液 故障检测与诊断 聚类分析 主成分分析 |
DOI: |
投稿时间:2017-07-29
|
基金项目:空调设备及系统运行节能国家重点实验室开放基金项目(SKLACKF201606)。 |
|
Fault Diagnosis for Compressor Liquid Floodback based on PCA-Clustering |
Zhou Zhenxin,Li Shaobin,Tan Zehan,Chen Huanxin,Wang Jiangyu,Liu Jiangyan,Guo Yabin,Sun Shaobo |
(Department of Refrigeration & Cryogenics, Huazhong University of Science and Technology;Gree Electric Appliances, Inc. of Zhuhai) |
Abstract: |
The liquid floodback in a compressor has an adverse impact on the variable refrigerant volume (VRF) air-conditioning system, which will cause energy loss. Nowadays, Big Data is being broadly utilized in fault detection and diagnostic (FDD). Thus, the PCA-Clustering method, which is combined with Big Data, was proposed to diagnose compressor liquid refrigerant floodback fault. First, data quality was improved by data preprocessing; secondly, the principal component analysis (PCA) method was employed to obtain the new dimensional variable data; finally, the new principal variables were clustered to get the classification label of liquid refrigerant floodback fault data. The results show that the model can preferably diagnose the liquid refrigerant floodback problem in the absence of real label data, with the diagnostic rate of the compressor liquid refrigerant floodback reaching 94.29%. |
Key words: VRF air conditioning system compressor liquid refrigerant floodback FDD clustering analysis PCA |