基于重复滑模观测器的直线电机系统干扰估计
Repetitive sliding mode observer for disturbance estimation of a linear motor system

采用常规滑模观测器(SMO)进行干扰估计时, 对时变干扰无法实现无静差跟踪, 并且估计信号存在抖振。针对一种重复运动直线电机系统, 同时存在周期性和非周期性干扰, 该文提出一种重复滑模观测器(R-SMO), 能够有效克服常规SMO的不足。它将周期性干扰的动力学模型(即内模)引入到观测器中, 能够渐近收敛于周期性干扰; 而滑模切换项只需针对非周期性干扰进行设计, 可以大幅度削减抖振。该文给出了R-SMO的渐近收敛性证明。实验结果表明: R-SMO能够有效地估计出直线电机系统的干扰, 用于前馈补偿后提高了跟踪精度, 并且在收敛精度和抑制抖振方面均优于常规SMO。该方法还可以扩展到多周期干扰系统和非重复运动领域。
Abstract：The drawback of the conventional sliding mode observer (SMO) when used for disturbance estimation is that it leads to nonzero estimation errors for a time-varying disturbance and chattering. A repetitive SMO (R-SMO) is developed for a repetitive linear motor system where periodic and non-periodic disturbances exist at the same time, which overcomes the drawback of the conventional SMO. The R-SMO incorporates the internal mode of the periodic disturbance to achieve asymptotic convergence to the periodic disturbance. The switching term then only needs to estimate the non-periodic disturbance which significantly reduces the chattering. The R-SMO is shown to be asymptotically convergent. Tests show that the R-SMO well estimates the disturbances of a linear motor system, improves the tracking accuracy after feedforward compensation and is superior to the conventional SMO in terms of convergence accuracy and chattering. Extensions to systems with multiple periodic disturbances and non-repetitive systems are discussed.