ABSTRACT
This chapter introduces a method to design Smith predictors (SP) when the plant is not precisely known. The method includes two criteria. The first one, presented as first algorithm, is based on bandwidth frequency considerations. It finds the set of model candidates for the plant so that, if the SP adopts one of them, the desired bandwidth is not reduced by the effect of the parameter uncertainty. The second criterion, presented as second algorithm, introduces some guidelines to improve the design of the SP by using the QFT technique. The chapter provides an illustrative example to show in detail how to effectively apply the methodology step by step to a plant with time-delay and model uncertainty. It introduces a methodology to design an internal-model based controller when the plant is not precisely known. For open-loop stable systems, the tracking problem of the non-minimum phase systems can be improved substantially by introducing the right half plane zero compensation.
