Model-based iterative control design

For a plant with high frequency almost-resonant modes it is very difficult to get a reliable model to design the suitable control. High gains at low frequencies and the presence of noise mask their influence, which appears if large bandwidths are required. For that purpose, an iterative identification/control-design technique is proposed. Two different control design approaches are considered: cancellation and generalised predictive controllers (GPC). The modelling techniques are very simple, step response and signal spectrum analysis. The results obtained for two academic examples illustrate the possibilities of this approach. The selection of the sampling period, in digital controllers, is also discussed.