Application of Fisher-PCA in Fault Diagnosis of TE Process

For the current plant big data, high-dimensional production environment, and the need for quick diagnosis of plant failures, this paper proposes an improved Fisher-PCA dimension reduction method for quick fault diagnosis of the entire system. This method is based on principal component analysis. Using the advantage of PCA dimensionality reduction on data loss, the Fisher criteria function is rapidly diagnosed in high-dimensional data. This method is applied to the TE process which is a real simulation of the strong coupling chemical production process. Simulation experiments show that this method can effectively distinguish the high-dimensional coupled fault data and normal data, and the method is practical and effective.