MODEL UPDATING BASED NONLINEAR JOINT IDENTIFICATION IN STRUCTURES UNDER RANDOM EXCITATION

Structural joints are almost nonlinear and their stiffness and damping vary versus amplitude of motion. Model updating in frequency domain can be used to capture the amplitude dependent behavior of stiffness and damping of joints. The amplitude of motion grows as the level of input force increases therefore, the joint's stiffness and damping varies. To determine this variation an underlying linear system which is obtained by ignoring the nonlinear elements form the model in simulation or exciting the system with very low excitation in practice, is considered as a basis. In a sequential increment of excitation and consequently response, model will be updated and calculated modification values will be joint properties at related response level. At the first step, underlying system is considered as basis and modified. For the next steps, when the response level increases, the last previous modified model is employed as basis. A Simulink model of a MDOF lumped system with frictional nonlinearity is considered to evaluate the proposed method. This system is subjected to a band limited white noise at various level of intensities, related Frequency Response Functions (FRFs) are extracted and model is updated at each level. Results are compared with Harmonic Balance Method (HBM). Good coincidence of the results with HBM, besides the confirmation of proposed method, expresses that joint's properties are similar in harmonic and random excitation. Also, it will be shown that proposed method is capable to determine the nonlinearities locations in the structure.