Characterization of optimal inputs by generating functions for linear system identification

Optimal input signals for Parameter estimation in linear dynamical systems may be characterized by generating functions in the form of linear difference equations. Results are derived for linear regression processes and for general linear systems. Discrete optimal control theory is applied to derive results for the latter case. The parameters of the difference equation may be considered as input information compression, and can be used to generate input sequence on-line during data acquisition for process parameter tracking.