带锚FRP受剪加固梁非剥离破坏模式BP网络预测

Prediction of failure modes for RC beams shear-strengthened with well-anchored FRP composites using neural networks

对带可靠锚固FRP受剪加固混凝土梁的非剥离剪切破坏模式做了细化分类,即包括FRP断裂控制的破坏、受压区混凝土(达到极限应力状态)压碎控制的破坏、FRP断裂与混凝土压碎同步发生的界限破坏等3种模式;利用BP神经网络建立了带锚纤维受剪加固梁破坏模式的智能预测模型,与31根非剥离破坏加固梁试验的对比结果显示:模型总体精度达到90%,说明建立的破坏模式网络预测模型适用于带锚纤维受剪加固梁非剥离剪切破坏模式的判别。

For RC beams strengthened in shear with well-anchored FRP strips,potential key factors causing shear failure are described which would give rise to three kinds of shear failure modes.A model of artificial neural network（ANN）has been developed to predict the shear modes of failure for them.And the prediction by the ANN model agrees well with the observation from the test reports.The result showed that the proposed model is fit for predicting the shear failure modes of beams shear-strengthened with wellanchored FRP composites.