System identification methods for dynamic structural models of electronic packages

In this paper we discuss two methodologies for system identification of structure models. Finite element software is coupled with optimization and estimation software to implement the techniques. Both methods update physical parameters in the finite element model and utilize iterative solution techniques. The first method integrates MSC/NASTRAN with a Bayes estimation software tool which is being developed for use with general finite element models. The second method utilizes a math programming optimization technique. Both methods were applied to a finite element model of an electronics package using both simulated and measured data. Results are presented using both eigenvalue and eigenvector data. The results showed the importance of using both eigenvalue and eigenvector data for the model of the electronics package. A discussion of the features of each method is included. 24 refs., 8 figs., 5 tabs.