Enhanced batch process monitoring using Kalman filter and multiway Kernel principal component analysis

  • Authors:

  • Yong-Sheng Qi;Pu Wang;Shun-Jie Fan;Xue-Jin Gao;Jun-Feng Jiang

  • Affiliations:

  • College of Electronic Information and Control Engineering, Beijing University of Technology, Beijing and College of Information Engineering, Inner Mongolia University of Technology, Huhhot Inner M ...;College of Electronic Information and Control Engineering, Beijing University of Technology, Beijing;Corporate Technology, Siemens Ltd., Beijing, China;College of Electronic Information and Control Engineering, Beijing University of Technology, Beijing;Corporate Technology, Siemens Ltd., Beijing, China

  • Venue:
  • CCDC'09 Proceedings of the 21st annual international conference on Chinese control and decision conference
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Batch processes are very important in most industries and are used to produce high-value-added products, which cause their monitoring and control to emerge as essential techniques. In this paper, a new method was developed based on Kalman filter(KF) and multiway kernel principal component analysis (MKPCA) for on-line batch process monitoring. Three-way batch data of normal batch process are unfolded batch-wise. Then KPCA is used to capture the nonlinear characteristics within normal batch processes and set up the more accurate monitoring model of batch processes. The on-line monitoring uses a Kalman filter which can estimate the entire trajectory of the current batch run. Comparison of the monitoring performance of the method with that of the traditional multiway principal component analysis (MPCA) method on a benchmark fed-batch penicillin fermentation process shows that the proposed method had better monitoring performance, and that fewer false alarms and small fault detection delay were obtained. In both off-line analysis and on-line batch monitoring, the proposed approach can effectively extract the nonlinear relationships among the process variables.