A Direct Adaptive Control Scheme for Disk File Servos

Abstract

This paper presents a direct adaptive control scheme for disk file track following servos. The scheme is designed to enhance the performance of a fixed observer based controller in the face of plant uncertainty and operating in a stochastic environment. The compensator used in each disk file is individually tuned with the use of an add-on adjustable FIR filter, Q, utilizing a linear fractional transformation (LFT) known as the Youla-Kucera parameterization. Similarly, the difference between the actual disk drive dynamics and the nominal model is characterized by an unknown transfer function, S, using the Youla-Kucera parameterization applied to the plant rather than to the compensator. The adaptive scheme utilizes two parameter adaptation algorithms running simultaneously. An recursive least squares (RLS) algorithm is used to identify the transfer function S which quantifies the mismatch between the actual disk file and the nominal disk file model. A second parameter adaptation algorithm is used to identify the finite impulse response (FIR) compensator Q. The goal of the overall adaptive controller is to minimize the specified H2 cost.