Damage identification via orthogonal empirical mode decomposition of curvature mode shapes

Time-frequency vibration signals processing technique and modal-based properties are combined to localize structural damage. The approach is devised to detect and locate damaged areas in the structure from its free vibrations and relies on the comparison between information gathered for the undamaged and damaged states. Extending previous studies based on non-stationary acceleration data, in this work strain data are considered. These are decomposed by applying the orthogonal empirical mode decomposition to derive the orthogonal intrinsic mode functions. The latter provide a multi-frequency basis enabling to build “pseudo modal curvatures” based on which a new damage index is proposed. The performance of the devised approach is shown through pseudo-experimental tests on a damaged arch.