基于BP神经网络的FRP加固混凝土柱承载力预测

BP Neural Network-Based Prediction of Load-Bearing Capacity of Concrete Column Reinforced by FRP

为提高纤维增强复合材料(FRP)加固混凝土轴压柱承载力的计算精度,建立了FRP加固混凝土轴压柱承载力的BP神经网络预测模型.利用大量试验数据对神经网络模型进行训练,并用训练成熟的神经网络模型对FRP加固混凝土轴压柱的承载力进行了预测.通过模型预测值与试验结果的比较,证明该模型的预测结果具有一定的可信度,最大误差不超过15%,比其他计算模型的精度高.

In order to enhance the calculation accuracy of concrete columns reinforced by FRP （fiber reinforced polymer） under axial compression, a BP （back propagation） neural network model was established to predict the load-bearing capacity of a concrete column reinforced by FRP under axial compression. The BP neural network model was trained by volume test data, and using the trained model, the load-bearing capacity of concrete columns reinforced by FRP was predicted. A comparison between the predicted and experimental results shows that the BP neural network model can consider more affecting factors and is reliable. Moreover, its maximum error is less than 15%, and its precision is higher than other models.