IEEE TRANSACTIONS ON AUTOMATIC CONTROL, v.65, no.2, pp.695 - 710
Abstract
The simultaneous presence of parameter variations, time-varying disturbances, and input delay of an n-th order dynamical system makes tracking control design difficult. Existing model-based control methods can handle only a subset of the above complexities and hence become insufficient. In this paper, a prediction-based adaptive robust control (PARC) framework is proposed for high performance control of systems subject to the above complexities, which involves: i) prediction-based projection type adaptation laws with model compensation to reduce the effect of parameter uncertainties under delay and time-varying disturbances; ii) a robust prediction scheme that factors in both unknown parameters and disturbance uncertainties under delay to handle the resulting unmatched disturbances; and iii) prediction-based continuous robust feedback to attenuate the cumulative effect of disturbance terms due to uncertain prediction. The controller guarantees semi-global, exponential convergence of the tracking error with an ultimate error bound proportional to delay and disturbance. The controller effectiveness is demonstrated with a flight control example and compared with the b.