Fault Detection and Isolation of Industrial Fermentation Process Based on Semi-supervised Convex Nonnegative Matrix Factorizations

Abstract

This study proposes a fault detection and isolation (FDI) approach based on a semi-supervised convex nonnegative matrix factorization (SCNMF) algorithm. In contrast to the existing nonnegative matrix factorization (NMF) algorithm, SCNMF uses the convex combination of each class of labeled samples to calculate the clustering centroid of the samples. The convex combination enhances the accuracy of the clustering centroid and improves the clustering performance of SCNMF. Moreover, the SCNMF-based FDI method is suitable for overcoming the challenge of conducting FDI with insufficient labeled samples. Using a case study on FDI for a penicillin fermentation process, the effectiveness of the SCNMF-based FDI method was validated.