Fault classification on Tennessee Eastman process: PCA and SVM

A problem focused on fault classification is studied in detail in this article. Two classification method, support vector machine and principal component analysis, are utilized to process this issue. Support vector machine, a common used binary classifier, is utilized as a multi-class classifier in this paper. There are several approaches to modify the binary classifier into multi-class classifier, and the "one against one" approach is chosen in this paper. Principal component analysis (abbreviated as PCA), regularly utilized to process interrelated variables and dimensionality reduction problems, is used as a fault classification algorithm in this essay. A simple comparison is made in the end of this article from the aspect of classification accuracy, and principal component analysis classifier shows a better classification performance.