Abstract
The deterioration of model uncertainties is an important factor restricting the control performance for hydraulic servo systems. This paper proposes a fuzzy-based adaptive model predictive control (MPC) method with a linear extended state observer (LESO) for such systems. The LESO is employed to estimate the total disturbances and other unmeasured states to predict system behavior. Based on the constructed objective function, the optimal output sequence is obtained via the receding horizon and repeating optimization. As for the output feedback MPC, the influence of key parameters on observation errors is analyzed and revealed in the face of deteriorating model uncertainties. Furthermore, its impact on the tracking performance of the system is considered according to the closed-loop state equation on the premise of asymptotic stability. Based on this, a fuzzy-based weight coefficient self-tuning strategy is developed to achieve the bounded observation errors. That is, during the repeating optimization, the defined and estimated disturbances are dynamically adjusted, simultaneously. The closed-loop tracking errors are thus improved compared with the case of the fixed weight coefficient. The simulation results indicate that the proposed fuzzy-based adaptive MPC controller shows an excellent tracking performance in the case of the deterioration of the model parameter uncertainty and the unmodeled uncertainty.
