Fault diagnosis model of batch process based on improved KFDA

For complex batch processes, it is possible to encounter the problem of singularity of kernel matrix during the calculation of kernel Fisher discriminatory analysis (KFDA) model. In this paper, an improved KFDA algorithm is proposed for fault diagnosis of nonlinear batch processes. Firstly, the original data is projected from the original space to high dimensional space by kernel functions. Secondly, in the calculation of KFDA, the orthogonal matrix is obtained by singular value decomposition for kernel within-class scatter degree matrix. Finally, the processed data and kernel within-class scatter degree matrix is projected onto a nonsingular orthogonal matrix after the decomposition. The feasibility and efficiency of the proposed method is demonstrated through beer fermentation process. (C) 2017, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.