摘要
针对空调系统运行过程具有非线性和动态特性的特点,提出了一种基于动态核主元分析的传感器故障检测方法。该方法采用核主元分析提取系统中的非线性冗余信息,建立核主元模型,再引入指数加权的定义,进行在线诊断的同时对模型进行实时更新,得到了改进的动态核主元模型。选择SPE统计量作为系统是否发生故障的依据。最后通过贡献图法实现了对故障变量的分离。将此方法应用于某地源热泵系统的传感器故障检测,结果表明,该方法能够实时更新核主元模型和置信限,成功分离故障变量,且和传统主元分析法相比具有更好的故障诊断效果。
In view of the process is nonlinear and dynamic in air-conditioning system,a fault diagnosis method based on dynamic kernel principal component analysis is presented.The model was constructed by kernel principal component analysis for extracting the system’s non-linear redundant information.Then the definition of exponentially weighted was introduced for renew the model when it diagnose online.So the improved dynamic KPCA model is proposed.The fault can be detected by the squared prediction error(SPE).At last,the faulty variables was isolated by contribution plots.The method has been applied in the sensor fault diagnosis for a ground-source heat pump system,and result show that the method can renew the KPCA method and control limit in real time,isolated the faulty variables successful,and it has better fault diagnosis effect compare with traditional principal component analysis.
作者
苗雨阳
刘成刚
MIAO Yu-yang;LIU Cheng-gang(School of Environment Science and Engineering,Suzhou University of Science and Technology,Jiangsu Suzhou215009,China)
出处
《机械设计与制造》
北大核心
2020年第7期161-164,169,共5页
Machinery Design & Manufacture
关键词
核主元分析
空调系统
传感器
故障检测
贡献图
Kernel Principal Component Analysis
Air-Conditioning System
Sensor
Fault Diagnosis
Contribution Plot