【摘要】 考虑螺旋肋FRP筋与混凝土粘结强度相关影响因素并搜集参考文献建立数据库。数据库被随机分成两个数据集，其中64组数据用于训练，17组数据用于预测，采用多层感知机算法构建了相应的界面粘结强度预测模型并与基于试验数据拟合得到的数学公式进行了横向对比。结果表明：文中得到的基于多层感知机的粘结强度预测模型决定系数R~2=0.961 3,精度较高，有良好的性能；相比传统数学公式，基于多层感知机的模型精度评估指标均为最优。更多还原

【Abstract】 Considering the influencing factors related to the bond strength between spiral rib FRP bars and concrete and collecting references to establish a database.The database was randomly divided into two data sets, of which 64 sets of data were used for training and 17 sets of data were used for prediction.The corresponding interface bond strength prediction model was constructed by using the multi-layer perceptron algorithm and compared with the mathematical formula obtained by fitting the experimental data.The results show that the determination coefficient of the bond strength prediction model based on multi-layer perceptron is R~2=0.961 3,which has high precision and good performance.Compared with the traditional mathematical formula, the model accuracy evaluation index based on multi-layer perceptron is the best.更多还原