Experimental study on sensor fault detection and diagnosis and estimation of centrifugal chiller system

Abstract

The reliability and accuracy of sensor measurements are greatly of significance to control and optimal operation of chiller system,as well as the basis of fault detection and diagnosis of chiller component faults.This paper presents a principle component analysis(PCA)-based sensor fault detection and diagnosis and estimation method for chiller system.A PCA model is composed of the observations of correlated variables in centrifugal chiller system in normal operation conditions aiming to capture the systematic variations of chiller system. These variations are used for the fault detection and diagnosis of new observations in terms of Q-statistic and Q-contribution plot.This PCA-based chiller sensor fault detection and diagnosis method was validated using the experimental test data of a centrifugal chiller.