Decentralized Robust Control of Two Inputs - Two Outputs System

This paper deals with methods of decentralized control, for systems with two inputs and two outputs. Robust control methods were used for tuning of the decentralized controllers. The method considers diagonal transfer function as a nominal model and the coupling part as a model uncertainty. The set of possible controller parameters creates family of models, for which a robust controller was designed. Range of the possible controller parameters was derived on the basis of Palmor’s method of decentralized control. The ultimate gain of the diagonal transfer functions may be taken as maximum values for all possible controllers and difference between ultimate frequencies gives the range all possible ultimate frequencies used for controllers design. Approximation models are obtained experimentally by two relay autotuning experiments. Controllers are designed on the basis of common condition for robust stability and robust performance. The method was tested in MATLAB/Simulink first for accurate model and then for approximate model. The parameters of controllers as well as the time responses of approximate models did not too differ from accurate model. Advantage this method is in shorter experimental work in comparison with the classic Palmor’s method, as only two relay experiment is claimed instead of several iterations. The results of simulation show, that responses are slower with smaller overshoots. The work has been supported by the grant agency MSM 002 162 7505. This support is very gratefully acknowledged.