Fault diagnosis of joint Bayesian method and LDA feature extraction in complicated industrial process

Bayesian method is a class of data-driven fault diagnosis method which is a topic of significant practical interest. In order to improve diagnosis accuracy and reduce computation load, linear discriminant analysis (LDA) is employed to extract features before performing Bayesian diagnosis. It can maximize the explicit function to achieve the goal that within-class data points as close as possible and between-class data points as far as possible. Tennessee Eastman Challenge (TE) is utilized to verify the effectiveness of the proposed method.