A common issue in structural damage identification is the measurement information being limited while the baseline structural model is complex, which renders the inverse analysis underdetermined. Leveraging that the damage in its early stage usually affects only small area(s), a multiobjective optimization can be formulated by minimizing the difference between model prediction and physical measurement and concurrently enforcing the sparsity in damage locations. The challenges, however, lie in handling the high-dimensional parametric space with multimodal objective functions as well as the computational cost brought by incorporating the norm minimization. In this research, we synthesize a hierarchical framework that searches potential damage locations first and then delves into accurate characterization of damage. In particular, to address the high-dimensional and multimodal challenges intensified by high-frequency measurement toward high-precision identification, we adopt the Group Lasso technique for sparsity induction and use the technique for order preference by similarity to ideal solution to realize optimization phase transition, aiming at improving the computational efficiency as well as the accuracy. The validation of our approach through a piezoelectric admittance sensing testbed underscores its potential for structural health monitoring practices by providing a robust, accurate, and computationally efficient approach for early-stage damage identification and assessment.
High-Frequency Active Interrogation for Structural Damage Identification Enabled by Intelligent Hierarchical Search
AIAA Journal ; 1-14
2025-01-01
Article (Journal)
Electronic Resource
English
Recent advances in active damage interrogation
AIAA | 2001
|Recent advances in active damage interrogation
Tema Archive | 2001
|AIAA-2001-1442 Recent Advances in Active Damage Interrogation
British Library Conference Proceedings | 2001
|CLoVER: An alterntive concept for damage interrogation in structural health monitoring systems
Online Contents | 2009
|