Anti-Saturation Extended State Observer for Speed Predictive Control of Servo System

In the servo system, input signals often easily reach a state of saturation, resulting in the discrepancies between the controller's output and the actual application requirements. To address this issue, an integrated scheme, which includes an AntiSaturation Extended State Observer (ASESO) and an Enhanced Generalized Predictive Controller (EGPC), is proposed in this paper. Firstly, the ASESO is designed to identify both the input coefficient and the lumped disturbance for the servo system. Leveraging these identification values, a feedforward compensation component is devised to mitigate the impact of external disturbance on the system model linearization. Furthermore, compared with the traditional ESO, the ASESO can autonomously determine whether real-time input coefficient updates are necessary based on constrained-saturation conditions. This capability ensures precise parameter estimation, even in scenarios involving the system saturation, thus meeting the controller's identification requirements. Subsequently, the EGPC, which optimizes a quadratic cost function over a defined “cost range”, can align with the identification parameters, so that the system output can follow with the reference value while adhering to additional constraints. Finally, the stability of the proposed scheme is verified by Lyapunov approach. Simulation and experimental results demonstrate the effectiveness of this scheme under saturation conditions.