Robust Adaptive Tracking Using Mixed Normalized/Unnormalized Estimation Errors

Parameter adjustment mechanism has an important role to obtain the smooth and fast responses in adaptive control systems. Using the normalized estimation error can improve the robustness properties of the adaptive system despite the perturbations, whereas by which the admissible tracking error and fast convergence may not be obtained necessarily. This paper concerns with the design of a parameter adjustment mechanism ensures that robust, fast and smooth convergence is obtained despite the disturbances and parameter variations. The algorithm is developed based on a variable normalizing gain to guarantee the convergence and then improved by combining with an unnormalized estimation approach to meet all the desired specifications. The proposed algorithm is then applied to model reference adaptive control (MRAC) scheme to ensure that robust tracking is obtained despite the perturbations. Simulation results show the capability of the proposed algorithm compared to the pure normalized or unnormalized approaches.