Stiffness Identification by a Substructural Approach in Frequency Domain

The concept of substructure representation is often employed to reduce the size of large and complex models for efficient structural analysis. In the context of structural system identification, this concept has also been used recently to improve numerical accuracy and convergence characteristics. Several research works have thus far demonstrated the effectiveness and advantages of substructural identification. But interface response measurements are usually required as input to the substructure to be identified in these methods. In this paper, a novel substructural method for stiffness identification is proposed with several advantages including eliminating the need of interface response measurement and even force measurement. The identification problem is converted into one of minimizing the difference between measured and predicted displacements at a check point in the substructure. Analytical formulation and numerical study are presented for both lumped mass system and continuous system, taking into consideration the effects of incomplete and noisy data.