大跨度斜拉桥可靠度评估方法研究

针对大型复杂桥梁结构极限状态方程一般难以显式表达的特点，提出了基于神经网络的大跨度斜拉桥可靠度评估方法．通过Latinhypercube抽样技术对随机参数进行抽样，应用大跨度斜拉桥非线性有限元进行分析．通过对随机抽样的样本数据进行训练，应用神经网络的非线性映射和泛化技术，对大跨度斜拉桥的极限状态方程进行数值模拟．通过极限状态方程对随机变量的偏导数，求解结构可靠指标的优化问题，计算大跨度斜拉桥的可靠指标．结果表明：对于隐式极限状态方程的大跨度斜拉桥可靠度评估问题，本文方法具有较高的计算精度和较好的计算效率；荷载布置方式、作用位置等对斜拉桥可靠指标有很大影响；计入3种几何非线性效应后斜拉桥偏于不安全，其中斜拉索垂度非线性效应的影响最为显著．

Reliability Evaluation of Long-span Cable-stayed Bridges

An approach evaluating the reliability of long-span cable-stayed bridges using artificial neural network （ANN） is proposed in the paper when implicit limit state functions are normally encountered in the complicated bridges. Random variables such as material properties, physical dimensions and loads are sampled by Latin hypercube sampling （LHS）. The ANN can be trained using a small set of numerical values obtained from the deterministic finite element analysis for long-span cable-stayed bridges and sample data mentioned above. The trained ANN can map the structural responses and random variables and the limit state functions of long-span cable-stayed bridges can be approximated using ANN. Then the values and partial derivatives of the implicit limit state functions can be calculated. So the reliability index of long-span cable-stayed bridges can be calculated by solving an optimization problem. A numerical example is given. The results show that the accuracy and efficiency of the proposed approach are validated. Arrangements along the bridges and position of traffic loads have significant influence on the reliability of long-span cable-stayed bridges. Long-span cable-stayed bridges become insecure when three geometric nonlinear effects are taken into account, in which the sa~ of the inclined cable exerts noticeable influence on the reliability of long-span cable-stayed bridges.