海水腐蚀梁抗剪强度及其智能分析

为了确定海水腐蚀钢筋混凝土梁斜截面承载力特性，用人工海水对15根钢筋混凝土梁进行加速腐蚀试验，得到海水腐蚀下钢筋混凝土梁斜截面受剪承栽力．随着腐蚀循环次数增加，承载力与构件参数及循环次数之间存在非线性关系，而采用常规的方法建立它们之间的关系，难以用数学公式来表达．利用神经网络方法，建立3层BP网络模型，以拟合不同条件下海水腐蚀钢筋混凝土梁受剪承载力的变化规律．根据试验训练样本数据，对该模型进行训练，建立BP网络模型系统．结果表明，基于模型系统的拟合曲线与试验曲线符合较好，能够反映承载力和腐蚀循环因素之间非线性变化关系，为腐蚀钢筋混凝土梁的抗剪强度预测提供参考．

Shear Strength and Intellective Analysis of Reinforced Concrete Beams Corroded by Seawater

A total of 15 reinforced concrete beams corroded by acceleration with manual seawater were tested to investigate their shear strengths of oblique section. Mathematics model was hard to be established by ordinary methods to describe the nonlinear relationships between shear strength and parameters of beams, and cyclic times. Based on neural network, a method was proposed to set up the fitting curve of shear strength on reinforced concrete beams corroded by seawater. A three layers BP neural network model was given considering four factors. The model was trained according to the test data. The results show that curves obtained by BP neural network model have preferable accuracy. So it could be used to predict the shear strengths of reinforced concrete beams corroded by seawater.