Abstract
A wavelet-based iterative learning control (WILC) scheme is presented in this article. For motion control system using conventional current cycle error (CCE) type ILC scheme, undesired vibration may be induced by nonlinear disturbances such as dead-zone, backlash, friction or other non-smooth nonlinearities and the amplitude of vibration will rapidly grow up during the process of iterations. In other words, the unlearnable dynamics of the motion control system will corrupt the control profile and cause instability during the iterative operation. To improve the learning behavior, wavelet transform is employed to extract the learnable dynamics from measured output signal before it can be used to update the control profile. The wavelet transform is adopted to decompose the original signal into many low-resolution signals that contain the learnable and unlearnable parts. The desired control profile is then compared with the learnable part of the transformed signal. Thus, the effect from unlearnable dynamics on the controlled system can be attenuated solely by a feedback controller design. Convergence analysis is also presented to provide theoretical background. A typical DC servo system and a belt-driven system are employed as the control target for experimental verification. Experimental results have shown a much-improved speed-tracking performance. Furthermore, in order to meet the requirements of simple hardware, fast rapid prototyping and cost down, we also design a wavelet based iterative learning control system servo-chip implemented on a signal field-programmable gate array (FPGA) system. The experiment results indicate that the WILC IC has shown a significant improvement in speed.