Data-Driven Multiobjective Controller Optimization for a Magnetically Levitated Nanopositioning System

Abstract

The performance achieved with traditional modelbased control system design approaches typically relies heavily upon accurate modeling of the motion dynamics. However, modeling the true dynamics of present-day increasingly complex systems can be an extremely challenging task; and the usually necessary practical approximations often renders the automation system to operate in a non-optimal condition. This problem can be greatly aggravated in the case of a multi-axis magneticallylevitated (maglev) nanopositioning system where the fully floating behavior and multi-axis coupling make extremely accurate identification of the motion dynamics largely impossible. On the other hand, in many related industrial automation applications, e.g., the scanning process with the maglev system, repetitive motions are involved which could generate a large amount of motion data under non-optimal conditions. These motion data essentially contain rich information; therefore, the possibility exists to develop an intelligent automation system to learn from these motion data, and to drive the system to operate towards optimality in a data-driven manner. Along this line then, this paper proposes a data-driven model-free controller optimization approach that learns from the past non-optimal motion data to iteratively improve the motion control performance. Specifically, a novel data-driven multi-objective optimization approach is proposed that is able to automatically estimate the gradient and Hessian purely based on the measured motion data; the multiobjective cost function is suitably designed to take into account both smooth and accurate trajectory tracking. In the work here, experiments are then conducted on the maglev nanopositioning system to demonstrate the effectiveness of the proposed method, and the results show rather clearly the practical appeal of our methodology for related complex robotic systems with no accurate model available.

Nonlinear‐disturbance‐observer‐based predictive control for trajectory tracking of planar motors

“…The trajectory tracking motion demands dynamic performance due to its time-varying tracking positioning. For planar motors, most control methods have been developed for high-precision point-to-point motion; and only several control methods have been exploited for high-precision trajectory tracking motion, such as the optimal proportional-integral-derivative (PID) control [6], data-driven model-free optimal control [7], hybrid control combining of adaptive robust control and iterative learning control [8], and model predictive control (MPC) [9]. It is found that MPC is an attractive and promising optimal control method due to its merits, such as simplicity in design and implementation, fast dynamic response, broad applicability, and strong ability to multiobjective constrained optimisation [10][11][12].…”

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

“…Recently, extensive novel feedback controller structures have been proposed and validated on various precision motion systems for performance enhancement [6]- [9]. For a fixed feedback controller structure, there is also a growing body of literature that concentrates on parameter tuning and optimization, which enables the feedback controller to achieve superior performance [10]- [12]. However, one-degree-of-freedom (1-DOF) feedback control is weak to achieve perfect tracking performance on account of its inherent dynamical lag characteristic.…”

Learning-Based Repetitive Precision Motion Control with Mismatch Compensation

“…However, the mismatch modeling approach has not been applied to the run-to-run control problem to augment the nominal model and improve controller performance. [13] develops a data-driven gradient estimation scheme for repetitive precision motion control, but the proposed framework does not have formal stability guarantees.…”