A non-parametric algorithm for time-dependent modal analysis of civil structures and infrastructures

Abstract

Vibration-based monitoring strategies have been demonstrated to be effective tools in providing – in nearly real-time – reliable information regarding the integrity of structures and infrastructure systems. However, commonly used methods for vibration analysis and modal identification are not able to capture the time variation of the modal properties during single acquisitions, hence they cannot perform dynamic identification in the presence of nonlinearities or non-stationary input excitations. To overcome this limitation, a novel non-parametric algorithm for automatic time-dependent modal analysis is hereby presented and discussed. This Enhanced Modal Identification for Long-term Integrity Assessment (EMILIA) algorithm can compute time-dependent estimations of the natural frequencies and mode shapes that can be critical to the early identification of hidden damage. The dynamic characterization of a beam-like structure in sound and damaged conditions is carried out for numerical validation purposes, allowing to evaluate the reliability of the proposed method over different scenarios and comparing its efficiency against traditional algorithms. Finally, further tests are conducted to analyse the sensitivity of the EMILIA algorithm to its main parameters and components.