Abstract:

There are four types of failure modes of reinforced concrete (RC) beam-column joints under lateral load: shear failure in the core zone of the joint, shear failure in the core zone of the joint after the column end yielding, shear failure in the core zone of the joint after the beam end yielding and beam end failure. The beam-column joints with different failure modes have different influences on the structural performance. Therefore, in performance-based seismic design of structures, accurately classifying the failure mode of the component is the key to determine the deformation performance limit of a component. According to recent research, there is no sharp line separating failure mode of beam-column joints from each other, and the variation law with single influence parameter is not obvious, so it is difficult to distinguish the corresponding interval of different failure modes through a single influence parameter. Based on Fisher transform and Bayes classification principle, this paper proposes a more accurate and convenient method to distinguish the failure pattern of beam-column joints considering multiple parameters. The algorithm first uses Fisher discriminance to find the maximum separation projection space between classes, and then projects the original sample to the maximum separation space to obtain new samples. On this basis, the new samples for better classification are deduced according to Bayes classification principle. In the classification of beam-column joints failure mode, a multi-parameter discriminant equation is established based on this classification method considering four parameters, axial compression ratio, shear compression ratio, concrete compressive strength and stirrup characteristic value, which achieves the purpose of dividing the failure modes of beam-column joints and clarifying the corresponding intervals of different failure modes better.

Key words: reinforced concrete, beam-column joints, Fisher transform, Bayes classification, failure mode