(322f) A Multi-Model Recursive Subspace Identification Based Dynamic Soft Sensor Approach for Digester Control

Kraft pulping is the commonly applied chemical pulping process that normally utilizes a (Kamyr) continuous digester. The most important quality variable in pulping is the Kappa number, which represents the residual lignin content of pulp. In Kraft pulping, it is desired to minimize the variations of the Kappa number of the pulp product. The continuous digester control problem is very challenging [1-6], and the main reason is that the Kappa number is measured infrequently and not in real time. Additional challenges include unmeasured disturbances, long time delays, and process nonlinearity [3]. In our previous work [10], we proposed a new dynamic soft sensor approach based on partial least squares (PLS). In addition we derived the equivalency between the proposed PLS based soft sensor and the subspace identification based soft sensor [7-9], and we show that by reducing the future horizon in subspace identification to one (i.e. f = 1), the subspace identification method can be applied to closed loop data. The proposed soft sensor makes use of process inputs and secondary measurements such as effective alkali (EA), active alkali (AA), total titratable alkali (TTA) and total dissolved solids (TDS) at different locations of the digester. To cope with time-varying parameters such as woodchip composition and white liquor concentration, in this work; we investigate different adaptive approaches to recursively update the soft sensor model on-line. Due to ?messy? composition in a digester, industrial data often exhibit substantial noise and outlier measurements. On the other hand; industrial pulping processes suffers frequent missing data due to sensor failures. To address sensor failures, we propose a multi-model approach where multiple models based on different sensor inputs are running in parallel. The impact of different sensor faults on the Kappa number prediction is studied and different schemes of model switch are investigated using a simulator (extended Purdue model) [3] and an industrial digester from MeadWestvaco Corporation with a Duralyzer-NIR? digester analyzer system developed by Hodges et al. [1, 2] measuring liquid concentrations at different locations of the digester..

Key words: digester control, subspace identification, soft sensor, multi-model, recursive, near-infrared spectroscopy

References: 1. Hodges, R.E., ?Applications of near infrared spectroscopy in the pulp and paper industry?, Ph.D. thesis, Auburn University Press, Auburn, AL, 1999 2. Hodges, R.E. and Krishnagopalan, G.A., ?Near-infrared spectroscopy for on-line analysis of white and green liquors?, TAPPI J., Vol. 82(9), 101, 1999 3. Wisnewski, P., Doyle III F.J. and Kayihan F., ?A fundamental continuous pulp digester model for simulation and control?, AIChE J., Vol. 43(12) p3175-3193, 1997 4. Wisnewski, P.A. and Doyle III, F.J. ?Control structure selection and model predictive control of the Weyerhaeuser digester problem?. J. Proc. Control, Vol. 8, 487-495, 1998 5. Wisnewski, P.A. and Doyle III, F.J., ?Model-based predictive control for a continuous pulp digester?, IEEE Trans. Control Syst. Tech. Vol. 9, 435?444, 2001 6. Lee, J.H. and Datta, A.K., ?Nonlinear inferential control of pulp digesters?, AIChE J. Vol. 40, 50?64, 1994 7. J. Wang & S.J. Qin, Closed-loop subspace identification using the parity space, Automatica, vol. 42, 315-320, 2006. 8. P. Van Overschee & B. De Moor, N4SID: subspace algorithms for the identification of combined deterministic-stochastic systems. Automatica, vol. 30, 75-93, 1994 9. M. Verhaegen, Identification of the deterministic part of MIMo state space models given in innovations from input-output data. Aubomatica, Vol. 30: 61-74, 1994. 10. Galicia H.; He Q.P.; Wang J. A Subspace Identification Based Dynamic Soft Sensor Approach for Digester Control, AIChE annual meeting, 2008, Philadelphia, PA