Robust Design for Generalized Minimum Variance Controllers

In order to improve the robustness and noise resistance of generalized minimum valance control systems, several generalized minimum variance control schemes are synthetically analyzed. The output variance caused by stochastic noise is decomposed to two parts. One part accords with the output variance of minboum vedance control and the other is the additional term of output variance caused by the control weighting factors. At the same time, the sensitivity function of modeling error is also deduced. A new robust design method that can minimize the sensitivity and the additional part of output variance is Presented by regulating variable parameters of contollers. The simulation results of self-tuning control show the effect of this method.