摘要
软体执行器是一种强非线性和无限自由度的系统,对其进行精确的建模和控制存在着巨大的挑战。基于Koopman算子的数据驱动建模方法能为软体执行器建立一个面向控制的近似线性模型。但所建立的Koopman线性模型存在建模误差且会影响控制器的控制性能。为此,设计了一种基于Koopman算子的数据驱动建模方法,并将扩展状态观测器(ESO)和模型预测控制器(MPC)相结合设计了控制器(K-EMPC)。首先,利用在实验中获取的输入输出数据对软体执行器建立Koopman线性模型。随后,利用ESO在线实时估计系统模型的建模误差,并在模型预测控制器(MPC)中进行补偿,从而提高控制器的精度和鲁棒性。
Soft actuators represent highly nonlinear systems with infinite degrees of freedom,posing significant challenges for accurate modeling and control.The data-driven modeling approach based on the Koopman operator offers a framework for constructing an approximate linear model suitable for control purposes.However,the Koopman linear model is prone to modeling errors,which can degrade the performance of controllers.To address this issue,this paper proposes a data-driven Koopman operator-based modeling method and integrates it with an extended state observer(ESO)and model predictive control(MPC)to design a controller(K-EMPC).First,a Koopman linear model for the soft actuator is derived using input-output data from experiments.Then,the ESO estimates the modeling errors of the system in real-time,which are compensated for in the MPC,enhancing the controller’s accuracy and robustness.Compared to traditional Koopman model-based MPC controllers,the proposed K-EMPC strategy demonstrates superior control performance under modeling errors and uncertain disturbances.Finally,simulation results validate the effectiveness and advantages of the proposed control strategy.
作者
王雄壮
WANG Xiongzhuang(School of Mechanical and Electronic Engineering,Wuhan University of Technology,Wuhan 430070,China)
