预应力CFRP加固RC梁承载力的神经网络预测

Neural Network Prediction of Bearing Capacity of RC Beams Strengthened with Prestressed CFRP Sheets

基于人工神经网络的原理,建立多因数智能分析模型,对已有的数据进行反向传播(BP)神经网络的训练.利用训练成熟的神经网络,对承载力进行预测,分析混凝土强度、截面高度、配筋率、配布率,以及预应力度等参数对承载力的影响.结果表明,用训练好的网络模型可以较好地预测预应力碳纤维布(CFRP)加固后钢筋混凝土(RC)梁的受弯承载力,预测精度较高,可以处理模糊的、非线性的问题.此外,混凝土强度、梁截面高度、受拉钢筋配筋率和碳纤维布的配布率的增加,受弯承载力也相应提高,但施加的预应力提高对极限承载力并没有多大帮助,只提高构件的开裂荷载和屈服荷载.

Based on the principle of back propagation （BP） neural network, a multi-factors intelligence model to predict the bearing capacity of reinforced concrete （RC） beams strengthened with prestressed carbon fiber reinforced polymer （CFRP） sheets is established. By the model, the bearing capacity of RC beams strengthened with prestressed CFRP sheets is predicted, and the action of key parameters is analyzed, such as the concrete strength, height of section, the area ratio of steel bar to concrete, the area ratio of CFRP sheet to concrete, and, the prestress degree and so on. The result indicates that the model can better predict the bending bearing capacity of RC beam strengthened with prestressed CPRP sheets. With increasing concrete strength, height of beam section, the area ratio of tensile steel bar to concrete and area ratio of CFRP sheet to concrete, the bending bearing capacity increases. The increase of prestress enhances the crack load and yield load, but slightly improves the ultimate bearing capacity.