Principal component analysis (PCA) is an effective approach to process monitoring and quality control.Although extensive researches related with PCA-based process monitoring approaches have been reported

the characteristics and fault detecting behavior of PCA are still equivocal.The commonly accepted conclusions in this field often conflict with the root cause of process malfunction and lead to incorrect understanding of the detection results.The expectations of T 2 and SPE statistics are studied in this paper and their relations to the statistical parameters of process data are presented.These relationships reveal the influence factors of the T 2 and SPE tests and give a definite description of the detection behavior of PCA.Based on these relationships process disturbances and faults can be distinguished

which make further fault diagnosis more reliable.The acquired results are illustrated and verified by monitoring of a simulated double-effective evaporator.