Order-Reduced Modeling-Based Multi-Level Damage Identification Using Piezoelectric Impedance Measurement

Abstract

Identifying small-sized damage in early state is an important subject in structural health monitoring (SHM) field. Despite the success of inverse model updating analysis for damage identification practice, the critical challenge remains. The capture of small-sized damage requires high-resolution finite element (FE) model, resulting in a high-dimensional model updating problem without the prior knowledge. This is computationally intractable. In this research, we develop a multi-level scheme to conduct damage identification, which circumvents the curse of dimensionality. To facilitate the damage identification, the piezoelectric impedance that is sensitive to the small-sized damage is utilized and a large-scale electromechanical FE model is constructed accordingly to predict the impedance response. Based upon the full-scale FE model, an order-reduced modeling approach is developed and integrated into the model updating framework to expedite the analysis. Multi-objective simulated annealing (MOSA) optimization algorithm is employed to direct the model updating through minimizing the discrepancy between the impedance prediction and measurement. The case studies are carried out to illustrate the effectiveness of proposed methodology.