HUSSAIN, ALTHAF2026-08-132026-08-132026http://nits.ndl.gov.in/handle/123456789/76The present study investigates model-based vibration damage identification through an inverse finite element (FE) updating framework. Structural damage is characterized as a reduction in elemental stiffness parameters and identified by minimizing discrepancies between measured and numerically simulated dynamic responses. The inverse problem is formulated by updating stiffness-related properties to match modal and frequency-domain characteristics of the structure. A progressive modeling strategy was used to assess the performance and robustness of the inverse algorithms as the structural fidelity increases: initially using an Euler–Bernoulli beam model to assess optimization performance for bending-dominated behavior, extending the formulation to Timoshenko beam theory to account for shear deformation and rotary inertia effects for improved dynamic accuracy, and finally generalizing the methodology to a two-dimensional frame model to capture axial–bending interaction and geometric transformation effects for assessing damage localization in redundant structural systems. The damage identification problem was generated as a nonlinear optimization problem and solved using global metaheuristic algorithms, namely Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Simulated Annealing (SA), Teaching–Learning Based Optimization (TLBO), and Differential Evolution (DE). Three objective functions based on the frequency residual, Modal Assurance Criterion (MAC), and Frequency Response Function (FRF) have been considered to study the sensitivity of various vibration measures. Damage characterization was considered at both the elemental and zonal levels, where a two-step zonal analysis was proposed to improve localization and prevent false positives. Noise perturbation was also considered for frequencies, MAC, and FRFs to test the effectiveness under practical uncertainties. To alleviate the high computational load posed by the repetitive FE analysis during each iteration of the optimization process, a surrogate model was incorporated within the methodology. A non-parametric mapping between the damage parameters and the structure’s response was established through structured sampling of damage configurations and GP-based regression modeling. The surrogate model substitutes the FE model in the evaluation of the objective function, thus providing fast predictions of modal properties with accuracyen