Multivariate Statistical Kernel PCA for Nonlinear Process Fault Diagnosis in Military Barracks

• DOI: 10.14257/ijhit.2016.9.1.17
• Publication Date: 2016-03-11T09:51:51Z
• Authors: Kaiwen Luo; Shenglin Li; Ren Deng; Wei Zhong; Hui Cai
• Keywords: Kernel (algebra)

Because of the nonlinear characteristics monitoring system in military barracks, traditional KPCA method either have low sensitivity or unable to detect fault quickly and accurately.In order make use higher-order statistics get more useful information meet requirements real-time diagnosis sensitivity, a new detection is proposed based on multivariate statistical kernel principal component analysis (MSKPCA), which combines statistic pattern framework (SPA) (KPCA).First, transformation function are conducted technology moving time window used.Then, PCA executed obtained from first step.Moreover, T^2 SPE control limits them calculated.Finally, simulations typical numerical example show that MSKPCA effective than terms diagnosis.