Multiple nonlinear parameter estimation using PI feedback control

Abstract

Nonlinear parameters often need to be estimated during the building of chemical process models. To accomplish this, many techniques are available. This paper discusses an alternative view to parameter estimation, where the concept of PI feedback control is used to estimate model parameters. The approach can be used for multiple parameter estimation problems if the parameters can be determined independently. In addition, similarities with Kalman filtering are illustrated. Both approaches are applied and compared in multiple parameter estimation problems. Furthermore, the PI controller approach is applied in parameter estimation for an experimental batch distillation setup. It is shown that the PI controller approach performs comparably to Kalman filtering, but is much easier to set up.