School of Civil and Transportation Engineering,Hebei University of Technology
Natural Science Foundation of Hebei Province
神经网络（ANN）模型作为土木工程领域中一种有效的方法能够用于解决复杂的问题。基于试验数据采用神经网络对钢筋混凝土剪力墙的抗剪承载力进行预测。收集120160个钢筋混凝土剪力墙在低周往复荷载下的试验数据，建立其数据库。选取100140个试验样本对ANN模型进行训练，20个试验样本进行测试验证。ANN1和ANN2有14个输入参数：混凝土抗压强度、剪跨比、轴压比、竖向钢筋强度、横向钢筋强度、墙体竖向分布钢筋配筋率、墙体水平分布钢筋配筋率、边缘构件纵向钢筋配筋率、边缘构件横向钢筋配筋率、边缘构件与截面面积比、截面高厚比、总截面面积、墙高和截面形状。其输入数据分别被归一化到区间[0, 1]和[0.1, 0.9]。两个模型的输出数据均为剪力。对比分析ANN模型预测的钢筋混凝土剪力墙抗剪承载力与采用规范GB50011和ACI318-14公式计算的抗剪承载力，神经网络模型能够精确地预测钢筋混凝土剪力墙的抗剪承载力，具有较好的预测和泛化能力。
In various areas of civil engineering, Artificial Neural Network (ANN) model is regarded as an effective method to solve complex problems. In this study, the shear bearing capacity of RC shear walls was predicted using ANN models based on the experimental data. A database on the performance of RC shear walls established from the results of 120 160 experimental data under cyclic loading was built. Moreover, 100140 experimental samples were chosen to train the ANN models, while 20 experimental samples were used for validation. There are had fourteen inputs, including concrete compressive strength, aspect ratio, axial compression ratio, vertical bar yield strength, horizontal bar yield strength, web vertical reinforcement ratio, web horizontal reinforcement ratio, boundary region vertical reinforcement ratio, boundary region horizontal reinforcement ratio, sectional area ratio, sectional heigth thickness ratio, total section area, wall height, and section shape. ANN1 and ANN2 were normalized in interval of [0, 1] and [0.1, 0.9], respectively. The shear force of the RC shear walls was the output data for both models. The predictions of the shear bearing capacity of the RC shear walls by ANN models and code methods from GB50011 and ACI318 were compared. The results reveal that the developed models exhibit better prediction and generalization capacity for RC shear walls.