Eigenvector Modification through Design Changes

This paper presents a technique for finding design changes which improve the eigenvectors of undamped dynamic systems for which a modal model exists. The designer specifies portions of the eigenvector to be altered and a set of design changes that are reaonable to make. The algorithm then locates the combination of changes that, for the least cost, provides the closest approximation to the specified eigenvector change. This is accomplished by minimizing a function which becomes smaller as (1) the norm of the vector of design changes becomes smaller and (2) the specified portions of the eigenvectors tend to take on the values defined by the designer. An example is included.