The efficiency of a novel identification method for structural damage assessment using the first vibration mode data

摘要

Three significant tuning components in structural finite element model updating including objective function, optimization algorithm, and updating variables have a drastic influence on the accuracy of structural damage location diagnosis and intensity prognosis. These three components require both physical concepts and trial-and-error approaches. To assess damage in a structure accurately, the common information of several modes of the structure is required. The availability of higher modes data in engineering structures with a high degree of freedom is a complex task or even not practical in real cases. This study intends to propose a versatile objective function based only on the first vibration frequency and mode shape data. A new hybrid criterion called "Relative Discrepancy Function (RDF)" is proposed which is composed of relative differences of natural frequency and mode shape vector. Hereupon, the efficiency of the proposed identification method is evaluated through five sets including different robust objective functions and meta-heuristic optimization algorithms. These five damage identification sets are composed of three objective functions (Normalized Modal Strain Energy, Modified Total Modal Assurance Criterion, and RDF) and three optimization algorithms (Imperialist Competitive Algorithm, TeachingLearning-Based Optimization algorithm, and the Most Valuable Player Algorithm (MVPA)). Subsequently, three truss and frame benchmark structures are assessed by means of five identification methods in single and multiple damage scenarios. It is observed that MVPA has both the fastest convergence rate and the lowest computational run time. Furthermore, the damage assessment results illustrate that when merely the first vibration mode data are used, the proposed identification method (RDF, MVPA) not only predicts the damage location properly, but also estimates the damage intensity successfully. (C) 2019 Elsevier Ltd. All rights reserved.