摘要
主元分析法可以应用于空调系统的传感器故障诊断,但是测量数据中隐含的噪声及系统的动态性影响了这一方法在故障诊断时的效果。本文提出采用小波变换的方法对测量数据进行分解,利用不含噪声及动态性的低频信号进行传感器的故障诊断,即基于小波变换的主元分析故障诊断法。该方法采用一大型离心式制冷机的实测运行数据进行验证,且同时与常规的主元分析法进行比较,结果说明基于小波变换的主元分析法可以提高故障诊断水平。
Principle component analysis (PCA) can be used for sensor fault detection, diagnosis and estimation (FDD&E) in HVAC systems. However, noises and dynamics embodied in the measured data usually deteriorate the performance of PCA for fault diagnosis. This paper presents a wavelet-PCA method for FDD&E. This method decomposes the measured data into approximation coefficients and detail coefficients, and uses approximation coefficients for PCA analysis because the coefficients do not contain noises and dynamics. This method was validated using the real operation data of a large-scale centrifugal chiller system. This method can improve the performance of sensor FDD&E by comparing with the conventional PCA method.
出处
《建筑科学》
北大核心
2007年第12期72-75,共4页
Building Science
关键词
小波变换
主元分析
传感器故障
故障诊断
wavelet transform
principal component analysis
sensor fault
fault detection and diagnosis