论文部分内容阅读
在分析5种具有不同输入变量的神经网络模型的基础上,建立了钢筋混凝土无腹筋梁抗剪强度的优化人工神经网络模型。该模型具有4个输入变量(混凝土抗拉强度、剪跨比、纵筋配筋率和截面有效高度)和一个输出变量(抗剪强度)。通过对数据的放缩处理,提高了网络训练效率。此外还对我国GB50010—2002规范、ACI318-08规范、Eurocode2、日本JSCE规范和加拿大CSA A23.3-04规范的无腹筋梁抗剪计算公式进行了对比研究。研究表明,神经网络模型具有较高计算精度,能够很好地预测无腹筋梁的抗剪强度。在各国规范公式中,CSA A23.3-04规范的计算结果与试验结果吻合很好,我国GB50010—2002规范、ACI318-08规范和Eurocode2公式计算结果的离散性较大。
Based on the analysis of five neural network models with different input variables, an optimized artificial neural network model of shear strength of reinforced concrete beams without web reinforcement is established. The model has four input variables (concrete tensile strength, shear span ratio, longitudinal reinforcement ratio and effective cross-section height) and an output variable (shear strength). Through the data scaling process, improve network training efficiency. In addition, the calculation formulas of non-reinforced web beams with GB50010-2002, ACI318-08, Eurocode2, Japan JSCE and Canadian CSA A23.3-04 are compared. The research shows that the neural network model has high computational accuracy and can well predict the shear strength of non-reinforced web beams. Among the national standard formulas, the calculation results of CSA A23.3-04 are in good agreement with the test results. The discrepancies in the calculation results of GB50010-2002, ACI318-08 and Eurocode2 are large.