论文部分内容阅读
基于神经网络,对剪切连接、剪滞效应、混凝土板裂缝不同情况下的型钢-混凝土组合桥正常使用状态下的挠度进行分析。建立3种神经网络,分别针对简支桥、两跨连续桥及三跨连续桥。神经网络的计算量几乎等同于忽略剪切连接、剪滞效应、混凝土板裂缝不同的简支梁分析。采用有限元分析软件ABAQUS,生成神经网络的训练和试验数据。神经网络对很多桥梁都有效,分析结果误差很小。基于改进的神经网络,给出封闭解。神经网络和封闭解可用于设计中的快速挠度分析。
Based on the neural network, the deflection of the steel-concrete composite bridge under normal use under shear connection, shear lag effect and the crack of concrete slab under different conditions is analyzed. Three kinds of neural networks are established, which are respectively for simple support bridge, two-span continuous bridge and three-span continuous bridge. The computational cost of neural network is almost the same as ignoring shear connection, shear lag effect and simply supported beam with different crack in concrete slab. Using finite element analysis software ABAQUS, neural network training and experimental data generated. Neural networks are effective for many bridges, and the error of analysis results is very small. Based on the improved neural network, the closed solution is given. Neural networks and closed solutions can be used for fast deflection analysis in design.