\abstract[Abstract]{Buckling-delayed shear-link dampers are used in seismic-resistant structures as passive devices that concentrate energy dissipation while limiting damage to the primary system. Their geometric optimization requires balancing high energy dissipation with strict control of local damage. Finite element models can accurately reproduce the nonlinear cyclic response of these devices and provide internal quantities such as damage indicators and local distortion, but their computational cost prevents their direct use within iterative optimization loops. This work proposes an adaptive surrogate-assisted optimization framework for buckling-delayed shear-link dampers. First, experimentally calibrated nonlinear finite element models are used to generate reference datasets for dampers with different geometric and mechanical configurations. Supervised learning models are initially evaluated, with support vector regression and Gaussian process regression consistently providing high predictive accuracy. This suggests that kernel-based, distance-dependent approximations are well suited to the problem, motivating the introduction of radial basis function surrogates as a computationally efficient alternative. The surrogate predictions are coupled with a differential evolution algorithm through a damage-aware objective function that controls local damage and uses dissipated energy as a performance criterion. Optimized geometries are finally re-evaluated with finite element simulations. When the surrogate error exceeds the adopted tolerances, the new simulation result is added to the dataset and the surrogate models are retrained. In addition, SHapley Additive exPlanations are employed to quantify the influence of window thickness on damage distribution, with particular emphasis on the response of the surrounding frame. The proposed framework provides an efficient damage-aware optimization of seismic energy dissipation devices.}
\abstract[Abstract]{Buckling-delayed shear-link dampers are used in seismic-resistant structures as passive devices that concentrate energy dissipation while limiting damage to the primary system. Their geometric optimization requires balancing high energy dissipation with strict control of local damage. Finite element models can accurately reproduce the nonlinear cyclic response of these devices and provide internal quantities such as damage indicators and local distortion, but their computational cost prevents their direct use within iterative optimization loops. This work proposes an adaptive surrogate-assisted optimization framework for buckling-delayed shear-link dampers. First, experimentally calibrated nonlinear finite element models are used to generate reference datasets for dampers with different geometric and mechanical configurations. Supervised learning models are initially evaluated, with support vector regression and Gaussian process regression consistently providing high predictive accuracy. This suggests that kernel-based, distance-dependent approximations are well suited to the problem, motivating the introduction of radial basis function surrogates as a computationally efficient alternative. The surrogate predictions are coupled with a differential evolution algorithm through a damage-aware objective function that controls local damage and uses dissipated energy as a performance criterion. Optimized geometries are finally re-evaluated with finite element simulations. When the surrogate error exceeds the adopted tolerances, the new simulation result is added to the dataset and the surrogate models are retrained. The proposed framework provides an efficient damage-aware optimization of seismic energy dissipation devices.}
\keywords{Buckling-delayed shear link, seismic energy dissipation, FEM validation, surrogate modelling, machine learning, radial basis functions, Differential Evolution, SHAP.}
\keywords{Buckling-delayed shear link, seismic energy dissipation, FEM validation, surrogate modelling, machine learning, radial basis functions, Differential Evolution.}
