Abstract

Info

Title:: Neural Network Model of Moment-rotation Relation in Semi-rigid Beam-column Joints Considering Floor Stiffness

Author(s):: LIU Jian; PAN Peng; LI Dong-lun; ZHOU Guan-gen; YU Zhi-wei; CHEN Yuan; School of Civil Engineering, Guangzhou University

Keywords:: floor stiffness; extended end-plate; semi-rigid beam-column joint; moment-rotation; neural network; simulation analysis

Abstract:: The shortage of existing classic moment-rotation model which was semi-rigid connections of beam-column joints for steel structure was reviewed. The nonlinear simulation analysis of semi-rigid joints for end-plate steel structure with floor stiffness and without floor stiffness were carried out using nonlinear finite element method, and the moment-rotation relation in semi-rigid joints was compared with that in semi-rigid classic momentrotation model. Finally, based on the neural network intelligent algorithm, the ten parameters neural network model which considered the effect of floor stiffness was established. The study results show that the floor can increase the stiffness of the beamcolumn joints, and reduce the relative rotation of joints which makes the curves for semi-rigid joints in practical engineering has large error with the existing classic semirigid joint model; the neural network model of semirigid joints for steel structure with floor stiffness effect can simulate the reality loading and transformation performance of semi-rigid beamcolumn joints; the neural network model has higher precision and computational efficiency, reliability, validity and practicability. The obtained results can provide references for the further study in other semi-rigid connections with floor stiffness and application in practical engineering.