The Generalized Predictive Controller (GPC) is extended to the systems with a generalized linear model structure which contains a number of choices of linear model structures. The Recursive Prediction Error Method (RPEM) is used to estimate the unknown parameters of the linear model structures to constitute a GPC self-tuner. Different linear model structures commonly used are compared and evaluated by applying them to the extended GPC self-tuner as well as to the special cases of the GPC, the GMV and MV self-tuners. The simulation results show how the choice of model structure affects the input-output behaviour of self-tuning controllers.
|Title of host publication||American Control Conference|
|Place of Publication||Pittsburgh, PA, USA|
|Publication status||Published - 1989|
|Event||1989 American Control Conference - Pittsburgh, PA, United States|
Duration: 21 Jun 1989 → 23 Jun 1989
|Conference||1989 American Control Conference|
|Period||21/06/1989 → 23/06/1989|