Identification of linear systems: a general framework for the theory

A general framework for the theory of identification of linear, time-invariant, stable and finite-dimensional systems is presented. The emphasis is on state space forms. The first part deals with parameter spaces and the relation between transfer functions and parameters. Full parametrizations, canonical forms, overlapping descriptions and local coordinates are considered. In the second part estimation for given dynamic specification is treated. Maximum likelihood type estimators and subspace identification procedures are discussed with an emphasis on asymptotic properties. The third part deals with order estimation