Multi-Input Identification Using Uncorrelated Signals and Its Application to Dual-Stage Hard Disk Drives

The identification of the linear dynamics of multi-input discrete-time systems is considered. The proposed technique simultaneously perturbs all the inputs using a set of periodic signals which are uncorrelated with one another. The technique has the advantages of short experimentation time, no recursive computation, and being able to provide a quantification of the effects of noise and nonlinear distortion. The proposed method is extended to identification beyond the Nyquist frequency, capitalizing on the periodic nature of the perturbation signals. Simulation results on a dual-stage hard disk drive illustrate the effectiveness of the proposed approach.