Structural Damage Identification of Offshore Jacket Platform Based on the Pseudofree Response Data

Due to the influences of the excitation and measurement noise, the misjudgments are inevitable when the coefficients of an autoregressive (AR) model in the time series combined with neural networks are used to identify the damage. So, in this study, the pseudofree response data are extracted from the acceleration responses data with the random decrement (RD) technique, the AR model is employed to fit the pseudofree response data, the Akaike information criterion (AIC) is used to estimate the order of AR model, the coefficients of the AR model will be changed if the structure is damaged, so we can determine if the structure is damaged according to the changes of the coefficients of the AR model. If the structure is damaged, the differences of the first 4 order AR coefficients pre and post damage are extracted and composed as damage sensitive vector which is put into back-propagation (BP) neural network to identify the damage location. The numerical model of a four-floor offshore jacket platform excited with white noise is used to testify the proposed damage identification method, the different amplitudes of the white noise excitation and the different level of the measurement noise are also considered. The simulation results show the proposed method is almost not affected by the changes of excitation amplitude, and when the noise level is no more than 5 percent, the damage location can be identified by the method correctly. The proposed method uses the acceleration responses to identify the damage directly, which is not dependent on the modal parameters (frequency, mode shape, damping), therefore, it is suitable for the on-line damage identification.