采用电化学腐蚀的方法加速梁内钢筋锈蚀,使用分级加载方案对受腐蚀钢筋混凝土梁进行了抗弯承载力试验研究,获得了20根不同锈蚀率的钢筋混凝土梁承载力极限值.使用现有的试验数据构成训练样本对,供支持向量机进行预学习.在机理建模方法中,基于线性核函数分别获得了分段式和整体式的回归公式;在黑箱建模方法中,基于径向基核函数获得了黑箱预测模型.通过将预测结果和本文试验值进行对比,验证所提出的基于支持向量机的受腐蚀混凝土梁承载能力预测方法有效性;并将预测结果与其它回归公式对比,表明所提出的预测方法具有预测精度高、适用范围广和预测策略多等优点.
An electrolyte corrosion technique was used to accelerate steel corrosion. Under the multi-stage loading schemes, bending test of corroded reinforced concrete beams was studied. The failure load of twenty RC beams with different corrosion rate was obtained. Current experimental data considerd as training sample was used to train the support vector machines. In mechanism modeling, piecewise regression formula and entire regression formula were obtained by using the linear kernel function, respectively. In balck-box modeling, prediction model of black box was obtained by using the radial basis function kernel function. The proposed prediction method was verified by comparing the mean squared errors of all the prediction values with the experimental results. It was shown that the proposed method has high prediction accuracy, extensive applicable range and many predictive strategies.