中文摘要：

在传统概率分析方法的基础上，引入神经网络一蒙特卡罗技术，提出了基于神经网络一蒙特卡罗法的钢筋混凝土截面能力概率分析方法．以桥墩的截面尺寸、箍筋间距、钢筋屈服强度、钢筋的弹性模量、混凝土抗压强度以及混凝土弹性模量作为随机变量．每个变量取7个水平进行正交设计，组合成49个工况进行截面能力分析，并以此作为样本建立神经网络．应用蒙特卡罗法产生大量随机参数，并输入到网络进行仿真，再对仿真结果进行数理统计得到相应的截面能力概率特性．实例验证表明，该方法提高了钢筋混凝土桥墩截面能力概率分析的效率．

英文摘要：

A probability analytical method for cross-section capability of RC pier is presented in this paper based on Radial Basis Function Neural Network （RBFNN） and Monte-Carlo （MC）. Sectional dimension, stirrup spacing, yield strength of steel, elastic ratio of steel, compression strength of concrete and elastic ratio of concrete are regarded as random variables. Firstly, an orthogonal design table is built with 7 levels for every parameter and 49 work cases are gained. Then corresponding cross-section capability is gained after every work condition analyzed. Secondly, a neural network is built up with training samples which are integrated by the Random variables and cross-section capability as a vector. Thousands of analysis results are simulated by the RBFNN after new random parameters gained by MC are input into. Finally, the characteristic of cross-section of RC pier is gained by tradi- tional statistic method. An example is presented, which proves that the method presented in this paper is very applicable and increases the efficiency of the probability characteristic analysis of cross-section capability of RC pier.