Numerical and graphical description of the information matrix in calibration experiments for state-space models.

ABSTRACT

This paper describes a mathematical tool for local identifiability analysis that can easily be applied to high-order state-space nonlinear systems and implemented in simulators with a discrete-time approach. The methodology is based on the recursive numerical evaluation of a reduced information matrix during the simulation of a calibration experiment and in the setting-up of a group of information parameters based on geometric interpretations of this matrix. As an example of application, the proposed methodology has been used in the study of an OUR batch test from the point of view of ASM No. 1 calibration.