Ranking of Effective Parameters in RC Column Shear Failure Using Machine Learning Methods

This paper investigates the importance of different parameters on the shear failure mechanism of columns in reinforced concrete buildings. A database comprising ۲۰۰,۰۰۰ samples was created using the Monte Carlo algorithm, considering ۱۰ parameters including: column cross-section, column longitudinal and transverse reinforcement, beam and column length, concrete compressive strength, reinforcement yield strength, axial load ratio, thickness of infilled wall and also infilled wall to column height ratio. A reinforced concrete moment frame considering flexural and shear behavior of columns is developed in OpenSEES, and has been verified by laboratory study. In the next step, the failure mechanism of the column is derived from pushover analysis for each sample. Then, using PCA, decision tree, and F-test machine learning methods, the importance of each of the ۱۰ parameters has been investigated. Finally, the parameters of the column transverse reinforcement and also infilled wall to column height ratio have been determined as the most important and effective parameters in column shear failure. The result of this paper is useful for developing models and criteria for rapid short column identification in seismic evaluation of existing buildings.

کلیدواژه ها:

Shear Failure ، Pushover Analysis ، OpenSEES ، Monte Carlo ، Machine Learning