A Predictive Bandwidth Extended State Observer With Gain Optimization for Model Predictive Speed Control of Electrical Drives

Abstract

In this study, a predictive bandwidth extended state observer (PB-ESO) with gain optimization is put forward to obtain high dynamic performance and parameters robustness for the model predictive speed control (MPSC) of electrical drives (EDs). The proposed observer preserves the primary feature of standard low bandwidth ESO in terms of strong suppression to measurement noise, while overcoming their main drawbacks, namely the “slow convergence” and “disturbance estimation distortion.” First, the conventional ESO-based MPSC scheme is presented, followed by an analysis of the disturbance estimation mechanism of the ESO. Second, the PB, based on the recursive least squares algorithm, is designed to accommodate to the frequency of fast time-varying disturbances in advance. A general mathematical formula for the gain optimization of even-order ESOs is then derived to enhance the estimation accuracy of periodic dynamic interference. In consequence, the gains of PB-ESO are optimized, where stability and fast convergence are ensured rigorously. Finally, a speed control strategy, combining the MPSC scheme and the proposed PB-ESO, is developed in a rational way. Comparative experimental results with existing methods demonstrate the enhanced robustness of the proposed control approach for ED across four representative operating conditions.