Seismic performance assessment of dual system incorporating MRF and BRBF using dynamic ensemble machine‐learning model

Abstract

Steel structures are widely used for their high strength‐to‐weight ratio, flexibility in architectural design, ease of prefabrication, and rapid construction, which make them a preferred choice for both low‐rise and high‐rise buildings. However, dual systems incorporating steel moment‐resisting frames (MRFs) with buckling‐restrained braced frames (BRBFs) are increasingly adopted in seismic design due to significant stiffness, strength, and ductility. In such systems, the MRF contributes to global deformation capacity and energy dissipation, while the BRBF provides substantial lateral stiffness and strength without the degradation associated with conventional braces. Accurately assessing the seismic performance of these dual systems requires consideration of their complex interaction and nonlinear behaviour under seismic loading, which can be computationally expensive when relying solely on detailed nonlinear time‐history analyses. However, the proposed dynamic ensemble machine‐learning (DE‐ML) model provides a promising approach to overcome this challenge. By integrating multiple base learner algorithms and combining their predictions, ensemble ML method can improve accuracy, robustness, and generalization compared to single models. The dynamic training of the ensemble ML model can further improve it for general data assessment. The results show that the proposed DE‐ML model can estimate the seismic performance of 4‐, and 5‐story MRF‐BRBF structures corresponding to 2% drift ratio with error values of 1.29 and 1.06, respectively, which is significantly less than the error values of those base learner algorithms.