摘要
分析了基于PCA-SVDD方法的冷水机组故障检测效率,结合PCA和SVDD方法的优点,提出了一种基于PCA-SVDD的冷水机组故障检测方法.通过PCA将正常数据所在的测量空间分解为主元子空间和残差子空间,取正常数据的残差子空间得分矩阵作为目标类数据建立SVDD模型,利用RP-1043中冷水机组实验数据验证故障检测性能,并与传统PCA和SVDD冷水机组故障检测结果进行对比.结果表明:PCA-SVDD方法可用于冷水机组故障检测,进一步提高了故障检测能力,且故障检测结果整体优于传统SVDD和PCA方法;用于冷水机组常见的故障检测,获得了较高的冷水机组故障检测效率.此方法有利于及早发现故障,减少损失,对小幅故障检测效率的提高尤为明显.
A PCA （principal component analysis ）‐SVDD （support vector data description ）‐based method was presented for chiller fault detection .First ,chiller fault‐free operating data were decom‐posed into two subspaces ,i .e .principle component subspace （PCs） and residual subspace （Rs） ac‐cording to the PCA method ,and then a SVDD model was developed based on the residual score matrix to detect chiller faults .Chiller experimental data from RP‐1043 were exploited to validate the PCA‐SVDD‐based method ,and the PCA‐SVDD‐based chiller fault detection results were compared with those obtained by conventional PCA‐based and SVDD‐based methods .Results show that the PCA‐SVDD‐based method can be used for chiller fault detection ,and most of its fault detection results are better than those of PCA‐based and SVDD‐based method ,especially on lower fault severity level , w hich is conducive to early fault detection and reduce the loss .
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2015年第8期119-122,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(51328602)
压缩机技术国家重点实验室(合肥通用机械研究院)开放基金资助项目
北京建筑大学供热供燃气通风及空调工程北京市重点实验室研究基金资助课题(NR2013K02)
关键词
冷水机组
故障检测
主元分析
支持向量数据描述
检测效率
chiller
fault detection
principal component analysis
support vector data description
detection efficiency
