Monitoring of transitions in multi-grade continuous processes

Modern process industries are adapting to the global competitiveness by equipping themselves to be flexible and producing products of different grades. This paper proposes an offline monitoring scheme which is suitable for such a dynamic environment where process transitions occur in a multi-grade continuous process. Modeling of every possible grade change is required for proper fault detection and isolation. The proposed technique uses Multiway Principal Component Analysis (MPCA) to eliminate the necessity to develop an individual model for monitoring every grade change performed in the process and is illustrated on an industrial case study of paper machine, a typical example of a production line where multiple grade changes occur naturally.