Research On Disturbance Suppression Of Permanent Magnet Linear Servo System Based On Iterative Learning Control

Iterative learning control (ILC) as a novel technology based on data-driven control, adjusts iteratively control input to track accurately the desired trajectory in a limited interval, aiming at the nature of repetitive control on a controlled system, using historical control input and output error information. The dissertation is based on the project of national natural science foundation "research on rejection techniques of high-precision permanent magnet linear servo system (51075281)", as a result of using permanent magnet linear synchronous motor (PMLSM) in linear servo system, for direct-drive characteristics and request of high position and track error precision, therefore the proposed scheme to design a novel ILC controller to ensure the robustness, high precision, high positioning and tracking performances of the system, on the basis of the feedback control, combining with feed forward control.In consideration of two-dimensional (2D) characteristics of iterative learning control, the segmented synthesis ILC for the2D system was investigated. On the basis of robust analysis and design for the2D system, a new phase-lead ILC scheme with input and output feedbacks was presented, which gave solutions to the problem of segmented compensation for linear motor disturbances. When stochastic disturbances are in dominant position, the ILC convergence is seriously influenced and the feedback control is strengthened. When repeated disturbances are in dominant position, ILC is strengthened. The effectiveness and good performances of the proposed ILC schemes were illustrated by PMLSM simulation on minimum thrust ripples along time and iterative axes. The proposed ILC strategy contrasts with feedback control on actual experiment of PMLSM. The results in real time show that position precision of PMLSM is effectively improved by the scheme.ILC can solve contradiction between tracking performance and system robustness by introducting a learning filter. Feed-forward compensation adopts shift phase ILC with two different band learning filters, to ensure the quality of learning convergence and control precision. Application of time window theory to ILC on servo system also is a more effective method, the proposed lead phase ILC with time window is introduced into causal two-band learning filter, on the basis of the feedback control, combining with feed forward control. Results in real time show that position precision of PMLSM is effectively improved by the scheme.The position axis and speed axis in the linear servo system are unified as a state axis (S), so that the main periodic disturbances in linear motor are suppressed by a compensator along the state axis. Compensation coefficients along the position and velocity axes are updated by the ILC algorithm. The algorithm is a communication bridge between adaptive control and iterative learning control, widening the application scope of ILC. Energy function of Lyapunov is simultaneously used in stability analysis of learning systems. Results in simulation and actual experiment show that position precision is effectively improved by the scheme.