A robust stability analysis on learning control for Hard Disk Drives

This paper is concerned with a learning control method to achieve high-speed and highly accurate head positioning for Hard Disk Drives (HDDs) with high areal densities. The learning method improves transitional motion in track seeking and achieves quick settling. From the robust stability analysis of the learning system, the use of a low pass filter to remove high frequency components from the learned feedforward input is also proposed. This provides much higher stability against model uncertainties than that in the original system without a low pass filter. Finally, it has been shown by experiment that the proposed learning algorithm with a low pass filter is able to make seek time distribution small while maintaining stability for HDDs with plant perturbation.