Mixed sensitivity robust adaptive control in a one degree of freedom feedback configuration

Abstract

An on-line solution for a mixed sensitivity, robust control problem is presented as a method of auto tuning of the control parameters. Linear stabilizable and detectable Multi-Input Multi-Output (MIMO) plants are considered. This creates a bond between robust and adaptive control, keeping the benefits such as low frequency disturbance and high frequency unstructured attenuation in the resultant control system. The unstructured uncertainty is represented by a feedback uncertainty model. The analytic expressions of the controller are written in terms of a state space of the plant and the controller belongs to the parameterization of all stabilizing controllers. No model reference is needed for the auto-tuning process. The stability of the Mixed Sensitivity criterion is analyzed by the Small Gain Theorem using the measure of the signals of interest. In a similar way, the Small Gain Theorem for non-linear systems is used to assure the stability of the overall system with the auto-tuning process.