Blind Identification of Polynomial Matrix Fraction for Disturbance Rejection

Abstract This paper proposes an approach to blind system identification for disturbance rejection control. We first identify plant dynamics with unknown disturbance, based on a well-known FIR (Finite Impulse Response) approximation technique for BSD (Blind Signal Deconvolution) in terms of independence of signals. Since we are more interested in the system itself than signal deconvolution, we adopt polynomial matrix fraction with a given degree structure, and then adjust the coefficients by projecting the FIR learning law to this parameter space of modest dimension. Finally we design feedback control for disturbance rejection via the identified system representation. Numerical simulation for a simple example is carried out to illustrate the proposed approach.