Techniques for Implementing Structural Model Identification Using Test Data

Structural system identification methods are analytical techniques for reconciling test data with analytical models. For system identification to become a practical tool for engineering analysis, the estimation techniques/ codes must communicate with finite element software packages, without intensive analyst intervention and supervision. This paper presents a technique used to integrate commercial software packages for finite element modeling, mathematical programming techniques, and general linear system analysis. Two examples of application of this software using measured data are presented. The examples consist of a truss structure in which the model form is well defined, and an electronics package whose model form is ill defined since it is difficult to model such structures with finite elements. A comparison of the resulting updated models with the experimental data shows significant improvement.