Time‐iteration‐domain integrated learning control for robust trajectory tracking and disturbance rejection: With application to a PMLSM

Abstract

Iterative learning control (ILC) has been widely used to improve motion performance when reference trajectories and external disturbances are strictly repetitive. However, the occurrence of non‐repetitive trajectories and disturbances would significantly deteriorate the performance of traditional ILC methods. To solve this problem, a time‐iteration‐domain integrated learning control (TIDLC) scheme is proposed for enhancing robustness against non‐repetitive trajectories and disturbances. The TIDLC scheme consists of an ILC term and a time‐domain compensator (TC) term. While the ILC term that performs learning in the iteration‐domain preserves excellent performance under repetitive tasks, the TC term that is updated in the time‐domain can compensate for the variations of trajectories and disturbances. Stability and performance analyses are discussed in both the time‐domain and iteration‐domain. Experimental results on a permanent magnet linear synchronous motor (PMLSM) positioning system verify the validity of the proposed scheme.

Comparative Study of Performance-Oriented Feedforward Compensation Strategies for Precision Mechatronic Motion Systems

“…For a repetitive trajectory tracking task, ILC derives the optimal compensation term by previous tracking errors so as to improve tracking accuracy. Generally, the compensation term of ILC can be injected into reference trajectory before the closed-loop system [37]- [39] or control input before the plant [40]- [42]. The former arrangement is Serial ILC (Fig.…”

“…Recently, an accuracy-oriented cascaded ILC (CILC) approach has been proposed to further reduce the residual error of standard ILC [41], [42]. To extend the scope of ILC applications, researchers have also contributed to enhancing extrapolation capability for non-repetitive trajectories [39] and robustness to accidental external disturbances [40]. However, ILC is merely suitable for repetitive trajectory tracking tasks and sensitive to trajectory changes.…”

Adaptive PID controller of permanent magnet linear synchronous motor based on particle swarm neural network

“…Because it does not depend on the model of the controlled object and has strong adaptability, the PID controller is the most widely used control strategy in current PMLSM servo systems, such as the current loop controller or speed loop controller based on PID [15,16]. To ensure that the motor has good dynamic operation characteristics under actual operating conditions, the PMLSM servo system will combine new control methods with a PID controller to improve the system performance, such as the fuzzy control strategy [17], sliding mode controller [18], and iterative learning control methods [19]. In the literature [20], an internal model control PID method based on a linear extended state observer model was proposed.…”

Iterative learning control applied to distributed‐order linear time invariant MIMO systems to achieve learnability

“…The ILC scheme has emerged as a very important tool due to its novel theoretical development and their applications; for example, in recent years, the ILC scheme has been combined with a MPC in order to address tracking trajectory problems (see reference Liang et al [1]), and its performance has been improved using model-free disturbance observers, which are commonly used in systems that require a precise motion, thus enhancing the robustness of the system (see Su [2]). Also, in Liu et al [3], to avoid the deterioration of the classical ILC scheme due to the witticism of non-repeated trajectories and disturbances, a TIDLC scheme is proposed for improving robustness. At the beginning, the experimental evidence showed that the P-Type Law works properly concerning nonlinear robot dynamics.…”