A self-tuning robust track-following control of sampled-data hard disk drive servo-system

This paper presents a self-tuning scheme based on the response surface method (RSM) to find the optimal finite impulse response (FIR) Youla parameter for an observer based state feedback track-following controller that minimizes the 3 times standard deviation of the position error signal in an HDD servo system. All the tested Youla parameters to construct the response surface were selected within a robust stable region which was defined by an artificial neural network (ANN) trained off-line. Such that the H_∞, bound of some sampled-data system channels could be kept during the response data collection. The experimental data show that such a self-tuning scheme could improve the position accuracy considerably in a short tuning time without any prior knowledge of the disturbance and noise, while robustly stabilizing the system