Optimization Method for Magnetic-Head Positioning Control System in HDD Against Unexpected Plant Perturbations by Using Data Mining Techniques

In this article, we propose a data-mining optimization method for a magnetic-head positioning control system in the hard disk drive (HDD) against unexpected plant perturbations, all without the necessity of dynamic models or frequency response knowledge of the control system. Our proposed method empowers us to fine-tune feedback controller parameters using a systematic process based on experimental classification outcomes, employing support vector machine (SVM) techniques. These parameter optimizations are determined through the identification of hyperplanes serving as decision boundaries between two classes: “success” and “failure” in terms of achieving the desired performance. Consequently, we can anticipate achieving the targeted performance levels when confronted with unexpected plant perturbations through the utilization of these optimized parameters. Case studies conducted on a benchmark HDD problem illustrate the effectiveness of our proposed method in achieving the desired performance when confronted with unforeseen plant perturbations.