The key concept in principal component analysis (PCA) is to reduce a high dimensional data volume into a lower dimensional space, where the low dimensional data con-tams most of the useful information/variance contained in the original data set. The projection axes are re-ferred to as principal components. As such, PCA has been widely used in industrial process control as a stan-lord technique for data analysis and process abnormality identification一9,13,22一z3].In terms of fault detection, a set of PCA components should be determined for the healthy data set and then fault detection can be performed by checking whether or not the new incoming data lies in the space spanned by the healthy principal components. PCA divides the whole observable space into a principal compo-vent subspace and a residual subspace, and then performs the FDD using C}-test and Hoteling T2 test. In this context, the statistics (SPE) used for FDD is given by:-The key concept in principal component analysis (PCA) is to reduce a high dimensional data volume into a lower dimensional space, where the low dimensional data con-tams most of the useful information/variance contained in the original data set.　 The projection　axes　are re-ferred to as principal components.　 As such,　PCA has been　widely used in industrial process control as a stan-lord technique for data analysis and process abnormality identification一9,13,22一z3].In terms of fault detection, a set of PCA components should be determined for the healthy data set and then fault detection can be performed by checking whether or not the new incoming data lies in the space spanned by the healthy principal components. PCA divides the whole observable space into a principal compo-vent subspace and a residual subspace, and then performs the FDD using C}-test and Hoteling T2 test. In this context, the statistics (SPE) used for FDD is given by: