The inverse dynamic design technique presented herein tunes a structur
al model to achieve a prespecified frequency response at certain criti
cal points of the structure. The technique can correct structural mode
ls based on vibration response test data and also can optimize structu
ral design. The technique works by partitioning the second-order matri
x equations of motion into assigned and auxiliary degrees of freedom a
nd then performs a parameter optimization to assign new frequency resp
onse data to the model. Design variable linking, symbolic or sparse ma
trix solutions, and closed-form gradients are used to streamline the c
omputations. The technique still requires a large amount of computatio
n, and thus is most applicable to small or moderate size dynamic model
s, such as coarse mesh prototype models, or models with repeated geome
try or symmetry. Two small order example problems illustrate the techn
ique. The first example completely corrects the stiffness matrix of a
corrupted analytical model using simulated measured (exact) frequency
response data. The second example flattens the first resonance peak of
a truss structure by assigning a specified partial frequency response
to the model.