A local and global statistics pattern analysis method and its application to process fault identification

摘要: Abstract Traditional principal component analysis (PCA) is a second-order method and lacks the ability to provide higher-order representations for data variables. Recently, statistics pattern (SPA) framework has been incorporated into PCA model make full use of various variables effectively. However, these methods omit local information, which also important process monitoring fault diagnosis. In this paper, global (LGSPA) method, integrates SPA locality preserving projections within PCA, proposed utilize preserve both information in observed data. For purpose detection, two indices are constructed based on LGSPA model. order identify variables, an improved reconstruction contribution (IRBC) plot locate The RBC original calculated with method. Based variables' statistics, new built simulation results simple six-variable system continuous stirred tank reactor demonstrate that diagnosis can effectively detect distinguish from normal