摘要
在已有文献有关试验的基础上,引入径向基(RBF)网络理论,提出了型钢高强混凝土柱抗剪承载力RBF神经网络预测方法。以混凝土强度等级、剪跨比、轴压比和配箍率为输入参数,混凝土柱的抗剪承载力为输出参数,建立精确RBF神经网络模型,以多组不同试验数据分别作为训练样本和检验样本,对网络进行训练和检验,并把仿真结果与采用非线性最小二乘法拟合公式的计算结果进行了比较。在文中所提方法的基础上,对型钢高强混凝土的抗剪承载力进行的参数分析结果表明,用训练成熟的RBF网络进行仿真,避免了诸多人为因素的影响,大大提高了结果的精度,使计算更加准确、高效。参数分析还表明,型钢高强混凝土柱的抗剪承载力随着混凝土强度、轴压比和配箍率的增大而增大,但随着剪跨比的增大而减小,并且剪跨比对柱的抗剪能力的影响最大,轴压比、混凝土的强度和配箍率则趋于同等重要影响程度。
Based on the literature and Radial Basis Function （RBF） neural network, an approach to evaluate the shear strength of steel reinforced high-strength concrete columns was presented. Grade of concrete strength, shear span ratio, axial compression ratio and stirrup ratio were regarded as input factors, and shear strength of the steel reinforced high-strength concrete column were considered output factor. A RBF neural network was built up and trained with training samples which were got from the test, and tested by other samples. As a result, a neural network was gained which reflects the highly-non-linear relationship between the input factors and output factor. By comparison of RBF neural network and the formula gained from the method of least squares fit, some influence of man-induced factor was avoided by RBF neural network which also increases the accuracy of the results. Finally, based on the method presented in this paper, parametric analysis was made for the shear strength of steel reinforced high-strength concrete columns. The analysis result indicates that the shear strength increases with the increasing of concrete strength, axial compression ratio and stirrup ratio, while decreases with the increasing of shear span ratio. The most sensitive factor is the shear span ratio, and the axial compression ratio, the concrete strength and the stirrup ratio are almost the same important.
出处
《结构工程师》
2008年第3期60-65,共6页
Structural Engineers
基金
国家自然科学基金项目(50778131)
国家科技支撑计划项目(2006BAG04B01)
交通部科技项目(200631882225)
关键词
型钢高强混凝土柱
抗剪性能
径向基网络
steel reinforced high-strength concrete column, shear performance, RBF neural network
