Robustness of fuzzy control and its application to a thermal plant

The paper deals with the robustness of fuzzy control. The complementary sensitivity function which in the case of the cancellation control corresponds to the desired closed-loop transfer function must be shaped in order to fulfil the basic equation of robust stability. With classical robust control this leads to poor performance. Fuzzy models are treated as a superset of linear models which are softly switched. It is supposed that every linear model contributes to the output according to the fulfilment grade vectors of membership functions on the universe of discourse. Under this supposition the multiplicative uncertainty of the model is reduced and a better control can be achieved. The proposed procedure is first illustrated on a simulated first order nonlinear process and then applied to a thermal plant—the laboratory scale heat exchanger.