Enhanced predictive functional control based on equivalent input disturbance and generalized extended state observer for nonlinear systems with fast dynamics

Abstract

An approach for improving tracking performance of the predictive functional control (PFC) based on equivalent input disturbance (EID) and generalized extended state observer (GESO) is presented in this paper. Control structure of the proposed method includes two parts, a PFC controller and a state observer, and both of them can be designed separately. The PFC controller is employed to optimize the tracking control of a nonlinear system and its tracking performance heavily depends on the accuracy of the predictive model. Whereas, various internal perturbations and external disturbances always make the predictive model significantly deviate from the nominal model. To address this problem, a state observer is introduced into the proposed approach. All of the above-mentioned uncertainties are regarded as a lumped disturbance, and the lumped disturbance is estimated by EID-GESO as well as the system states in an integrated manner. The estimated disturbance can be eliminated in a negative feedback loop and then, a relatively accurate predictive model for the PFC controller is offered. Closed-loop stability of the composite control is also provided. Compared with previously related work, the notable feature of the proposed design is that the observerbased PFC is extended to nonintegral chain systems subject to mismatched uncertainties and better performance is obtained. Finally, the proposed approach is applied to a missile longitudinal autopilot design, and a comparison with some prominent methods in the presence of significant uncertainties demonstrates the robustness and effectiveness of the proposed design.

Disturbance observer based model predictive control for a VTVL reusable launch vehicle

“…The predictive function control method [3] retains the three basic characteristics of MPC algorithm, namely prediction model, rolling optimization and feedback correction [4] .…”

“…The five basic elements of PFC are base functions, prediction model, reference track, feedback correction, and optimization performance index. The block diagram of the basic principle is shown in the Figure 1 [4] .…”