مجيد كاري

Every year, earthquakes pose significant life and financial risks in various countries around the world. One of the concerns of civil engineers is to reduce the effects of earthquake forces on structures. Conventional lateral load-bearing systems, by offering inelastic behavior and large deformations, absorb earthquake forces and undergo residual deformations, which leads to disruption in building services after an earthquake. A new type of structural system, called self-centering systems, has been created to increase ductility and to overcome the disadvantages of conventional systems that suffer little or no damage during an earthquake. The goal of these systems is to reduce damage while ensuring the safety of the occupants. In these systems, with their special mechanism, they keep the structural parts in the elastic range, and energy dissipation and damage are done by repairable or replaceable friction elements at a low cost. Given the complexity of self-centering structures and the high computational cost of time history analyses, the use of modern methods such as machine learning helps us predict the behavior of centering systems in less time and with appropriate accuracy. In this research, after the actual modeling of centering structures, we proceeded to construct simplified models and, by providing modification factors, achieved accuracies equivalent to actual modeling. By using this modeling, we analyzed and created three datasets of 15,000 samples with different coefficients of variation, for a total of 45,000 samples., and total of 45,000 samples. Then we began training and testing various machine learning models and examined these models. In the end, the trained machine learning models were eva‎luated by a number of actual modeled structures and random parameters. Ultimately, it was observed that with appropriate accuracy, we can predict the roof drifts of the two core self-centering rocking steel braced systems.

سيستم مركزگرا , قاب فولادي مهاربندي , رفتار لرزه‌اي , يادگيري ماشين , پيش‌بيني

Self-Centering system , Steel braced frame , Seismic behavior , Machine learning , Prediction

Majid Kari

Vahid Broujerdian