为了对复杂结构进行随机分析及可靠度评估,提出一种复合算法。使用拉丁超立方抽样选取样本点拟合神经网络,基于显式化的神经网络解析式,采用皮尔森相关系数进行参数敏感性分析,利用一次二阶矩与重要抽样蒙特卡罗法进行可靠度评估,最后采用该算法对空间缆索自锚式悬索桥进行了静力随机分析及正常使用状态下的可靠度评估。计算结果表明:与大跨地锚式悬索桥静力受力行为不同,对于中等跨度自锚式悬索桥,主梁抗弯惯性矩是活载作用下主梁挠度最为敏感的因素;荷载的随机性是影响正常使用状态下结构可靠度的最敏感因素。
In order to perform complex structure stochastic analysis and reliability assessment,a new hybrid method was proposed,in which the Latin hypercube sampling technique was used to establish an artificial neural network model,and based on the explicit formula of the performance function from the well-trained artificial neural network model,Pearson linear correlation coefficient was adopted for sensitivity analysis and the first-order reliability method for the most probabilistic failure point and Monte-Carlo importance sampling technique for the estimation of the failure probability.The new hybrid method was applied to sensitivity and reliability analysis of Jiangxinzhou Bridge.The results show that different from large span earth-anchored suspension bridge,the bending inertia moment of the main girder is the most sensitive parameter to its deflection under extra load for the middle span self-anchored suspension bridge,and randomness of load is the most sensitive factor to structural reliability under normal service condition.