XGBoost 알고리즘 기반 철근콘크리트 기둥의 Backbone 곡선 파라미터 예측

초록

When assessing the seismic performance of reinforced concrete (RC) frames, it is important to use an accurate numerical model for columns because the seismic behavior of the columns significantly affects the structural performance. In most previous studies, the parameters of column models were determined using empirical equations. However, it is difficult to fully capture the complex and nonlinear characteristics of actual columns using empirical equations developed from regression analyses. This study developed a machine learning (ML) model to construct an idealized backbone curve for RC columns. For this purpose, test data for rectangular RC columns under cyclic loading were collected from previous research. Three damage states were defined to construct the backbone curve, and the accuracy of the proposed ML model was subsequently validated. It was shown that the measured backbone curves of the collected columns could be precisely predicted using the parameter values obtained from the proposed ML model.

키워드