The concepts and methodology of system identification and adaptive control in papermaking an presented and discussed. In particular the crucial problems of dead time compensation and proper interpretation of scanning gauge measurements are reviewed. Two new approaches developed recently in our laboratory to overcome these problems are presented.
The key points in extending the monovariable adaptive control schemes to the multivariable case are discussed. Successful applications of selftuning control to machine and cross-directional control of moisture and basis weight are outlined. The difficulties associated with multivariable adaptive control are illustrated by an example of colour control on a paper machine.
Finally, novel approaches to adaptive control such as genetic algorithms, neural networks, and multimodel techniques are presented. The promise of these recent advances in adaptive control is illustrated by an example of automatic compensation for species changes in kraft pulping.