A Bayesian approach to global mode shape identification using modal assurance criterion-based discrepancy model

Modal identification of large-scale engineering structures may require multiple measurements obtained from different locations, due to the limitations in the instrumentation equipment. In such cases, the global values for modal frequencies and damping ratios can be estimated over the measurement setups by simply obtaining their statistical expected values. Estimation of global mode shape vectors, however, arises as a much more challenging problem since the local mode shapes confine at different degrees of freedom of the measured structure. To overcome this problem, a novel Bayesian methodology is presented in this study using a modal assurance criterion-based discrepancy model. By making use of this model, a novel solution procedure is presented to the existing literature, and mode shape normalization-based problems arising in the estimation of covariance matrix are solved. The computational efficiency of the proposed method is verified by two numerical and one real data example. The conducted investigations reveal that the presented methodology shows a good performance in terms of computational accuracy.