On-line batch process monitoring using a consecutively updated hierarchical kernel partial least squares model

In the paper, a new approach, a consecutively updated hierarchical kernel partial least squares (UHKPLS) model was proposed. Using multiway partial least squares (MPLS) monitor industrial batch processes has followed disadvantages: 1) MPLS is a linear projection method, which cannot effectively capture the nonlinear features existing in most batch processes. 2) It is limited that complete batch process data is indispensable while the MPLS is applied in batch process monitoring. Hierarchical kernel partial least squares (HKPLS) can solve these problems. But HKPLS is a fixed-model monitoring technique, which gives false alarms when it is used to monitor real processes whose normal operation involves slow changes. So an on-line batch monitoring method that uses a consecutively updated hierarchical kernel partial least squares (UHKPLS) model was proposed to solve these problems. The proposed method was applied to monitoring fed-batch penicillin production. The simulation results clearly show that the ability of the proposed method which eliminates the many false alarms and provides a reliable monitoring chart.