Multivariable nonlinear control of a paper machine headbox

The nonlinear predictive control of a multivariable paper machine headbox is considered. The proposed approach allows the systematic design of a physical model-based control structure for the control of the process. This is achieved by embedding the nonlinear physical model of the process within the model predictive control approach. The use of physical models to motivate the controller structure and parameters moves away from the black-box approach to model-based control thereby allowing the use of prior system knowledge within the control design. The MPC approach was chosen because of its ability to handle constraints in a systematic way. The proposed controller is also related to decoupling and linearization of the plant based on a nonlinear feedback transformation. In the simulated closed-loop control of the headbox the effect of state and parameter uncertainty on the process operation is examined in detail