Simplified process monitoring method based on dynamic factor analysis and its application

Aimed at the dynamic characteristic of the process, this paper chooses suitable time lags for each variable by auto-correlation analysis and simplifies the augmented matrix to upgrade the result of monitoring and simplify the traditional dynamic process monitoring method such as dynamic principal component analysis (DPCA). Combining this thought with factor analysis (FA), this paper proposes a new dynamic factor analysis modeling method. The parameters are computed by expectation-maximization (EM) algorithm and the monitoring indices are constructed. This improved method is introduced into process monitoring and applied to the Tennessee-Eastman (TE) process. Compared with the traditional dynamic principal component analysis (DPCA) and dynamic factor analysis (DFA), the validity and superiority of the proposed method are shown by monitoring chart, missing detection rates and detection delays.