基于神经网络的十字形截面方钢管混凝土组合异形柱轴压承载力研究

STUDY ON AXIAL BEARING CAPACITY OF CRISSCROSS SECTION CONCRETE-FILLED SQUARE STEEL TUBE SPECIAL-SHAPED COMPOSITE COLUMN BASED ON THE NEURAL NETWORK METHOD

采用有限元法对方钢管混凝土组合异形柱轴压性能进行模拟分析,通过有限元模型的破坏形式和轴压承载力与试验结果对比,验证了有限元程序的可行性。基于有限元分析结果,训练神经网络,通过承载力的对比,二者误差较小,验证了神经网络的可行性和精确性。利用神经网络对十字形截面方钢管混凝土组合异形柱轴压承载力影响因素进行参数化分析,结果表明承载力随着钢材强度、混凝土强度、钢管高度、钢管尺寸和钢管厚度的增大而增大。

This study simulated the axial bearing capacity of concrete-filled square steel tube special-shaped composite column by FEM. Through the comparison test of the damage form and axial bearing capacity of FE model and the experimental results, the feasibility of finite element program was verified. Based on the FE results, the neural network was trained to get the bearing capacity. By comparing the result of those two bearing capacity , which indicated the error was small, the feasibility and accuracy of neural network had been verified. It had conducted parametric analysis to the axial bearing capacity of concrete-filled square steel tube special-shaped composite column by using neural network. The result confirmed that its axial bearing capacity increased as these factors increasing such as steel strength, concrete strength, the height, size and thickness of steel tube column.