Discrete-time model regulation control for systems with uncertain dynamics: Applications to electromechanical and electropneumatic systems.

摘要

A discrete-time Model Regulation approach for control systems with uncertain dynamics which includes model uncertainty and external disturbance is developed. The proposed approach regulates the Input/Output behavior of the actual control plant to its nominal model so that robust performance may be achieved regardless of the existence of uncertain dynamics. The actual implementation of the Model Regulation Control algorithm is achieved in discrete-time form by using inverse dynamics, time delayed signals and low-pass sensor filters. Causality, unacceptable zeros of the system and sensor noise are directly taken into consideration in the implementation of the control algorithm. Analysis on sensitivity reduction and disturbance rejection shows that the success of this approach is guaranteed in the low frequency range and effected by the delay time and the model uncertainty in the high frequency range. The stability of the compensated system is determined by the delay time and the model uncertainty. The critical design factor for this approach is to increase the sampling rate to extend the bandwidth for the acceptable sensitivity reduction and disturbance rejection and also to improve the stability margin. The Model Regulation Control is applied to a DC servomotor system for performing velocity, position and tracking control. Analytical studies and experimental results prove that both linear (payload variation) and nonlinear (friction force) uncertain dynamics are compensated by the Model Regulation Control. The application of the Model Regulation Control is studied. A linear model is experimentally derived at the chosen operating point as the nominal model for controller design. The uncertain dynamics considered are the state-dependent parameter variation and the unmodelled friction force. Experimental results verify that the pressure responses for different volumes and initial pressures can be controlled to be identical to the response of the nominal model. The position output is also controlled to behave linearly for motions covering almost the entire stroke.