Parameter identification methods for metamodeling simulations

A metamodel is a mathematical approximation of the system relationships defined by a high fidelity model or simulation. This paper presents methods that support new procedures that expanded the set of available metamodel representations be yond the traditional least squares formulation and added the capability to use dynamical metamodels. These methods compliment a new taxonomy of metamodel structures and procedures that separated the metamodeling process into a set of sequential decisions based on a priori information. This work was supported in part by The USAF Rome Laboratory Contract F30602-94-C-0110.