A new method for structural model updating and joint stiffness identification

A new model updating method is presented here with emphasis on identification of joint stiffnesses. A so-called reduced-order characteristic polynomial (ROCP) is first defined in terms of the measured natural frequency, the partial modal properties predicted from a flawed FEM model and the model parameters to be updated or identified. Based on the premise that each of the measured natural frequencies is a root of the ROCP, a set of non-linear equations becomes immediately available from which the interested model parameters can be determined. The experimental data required are simply the measured natural frequencies, rather than the usually incomplete and less-accurate displacement or FRF data. In addition, there is no modal matching needed here, which can lead to a considerable saving of the resources and efforts spent on measurements and correlation analyses. The proposed technique is verified by using simulation data. The current scheme is easily generalised into an advanced optimisation formulation, which would allow various constraints to be explicitly imposed upon the model parameters and a wide spectrum of algorithms to be directly used for finding a solution. (C) 2002 Elsevier Science Ltd.