Identifiability of linear parametric stochastic systems

Identifiability of a parametric stationarys system is defined by using only a description of the system and exogenous signals, observable and unobservable. For systems; described by linear difference input – output equations this definition leads to identifiability equations which reduce identifiability check to simple operations with system and model polynomials. Properties of models satisfying the identifiability equation are discussed. The identifying properties of the equation and of a method with a quadratic norm of residues are compared. The necessary and sufficient conditions for identifiability of systems of various structures are provided.