Subsystem State-Space Model Identification and Its Sensitivity to Test Variability

Paper in proceeding, 2004

A method for identification of a subsystem state-space model is presented. Having available models for the complete coupled system and for the remaining subsystem, the unknown model can be extracted via linear matrix operations only. Hence, the method is computationally cheap. However, as the a priori known models most likely are contaminated with modelling and parametric errors visavis physical reality, a sensitivity analysis is performed. Analytical expressions for the identified model sensitivity to parameterized uncertainties in the known models are derived. General conclusions based on inspection of the analytical expressions are validated by numerical experiments.