Reducing errors in the identification of structural joint parameters using error functions

A new method is proposed to identify the structural joint parameters directly from the frequency response functions (FRFs) of the substructures and the whole structure. The problem of the measurement noise with non-Gaussian distribution in the FRFs, and the problem of joints with very different orders of magnitude are especially discussed in this work. The new method uses an error function to select the best data to identify the individual parameter so that the new method can function well under different strict conditions. The accuracy of the new method and other two existing methods is compared under different conditions in this work.