Starting with uncertainty considerations of mathematical models, the c
oncepts of verification, validation, usability and falsification are d
efined. The tool for achieving a verified, validated and usable mathem
atical model is system identification, or more particular model updati
ng in the inverse formulation. Uncertainty modeling is performed for t
he mathematical model to be corrected by model decomposition (substruc
turing) and for the measured data by stochastic methods by applying es
timators. Problems in updating are discussed, as are the well-posednes
s of the formulation of the identification problem itself, the determi
nation of the model structure, and the order reduction of discretised
linear models. Numerical algorithms and regularisation methods are dis
cussed also. Applications are mentioned, such as the model-supported d
iagnosis of technical systems and the parameter estimation of constitu
tive equations. (C) 1998 Academic Press Limited.