湖南大学学报（自然科学版）2026，Vol.53Issue(7)：89-103,15.DOI:10.16339/j.cnki.hdxbzkb.2026063
基于生成对抗网络的FRP加固钢筋混凝土剪力墙滞回性能预测方法
Generative adversarial network method for hysteretic performance prediction of FRP-strengthened reinforced concrete shear walls
摘要
Abstract
Externally bonded fiber reinforced polymer(FRP)sheet is an effective way to improve the shear strength of reinforced concrete shear walls.Accurately predicting the hysteretic performance and skeleton curve can effectively improve the reliability of reinforced structure design and analysis.This paper proposes a prediction method for the hysteretic behavior and skeleton curves of FRP-strengthened reinforced concrete shear walls based on a generative adversarial network.By inputting design parameter information such as geometric dimensions,material strength,and load conditions of FRP-strengthened reinforced concrete shear walls,the hysteretic curves and skeleton curves are directly obtained.Eight sets of finite element simulation results under different working conditions were used to train the deep learning model,and two additional sets were used as the test set to evaluate model performance.The trained model was further applied to predict one set of real engineering case.The results show that it is accurate to use generative adversarial network to capture the characteristics of load and time and obtain hysteretic curves and skeleton curves.On the test set,the coefficient of determination(R2)reached 0.989 8,the root mean square error(ERMSE)was 15.630 kN,and the mean absolute percentage error(EMAPE)was 48.65%.The accuracy of this method was verified by comparing the prediction results of a variety of different deep learning networks and experimental results under the same working conditions.关键词
纤维增强材料/钢筋混凝土剪力墙/抗震性能/多元时序深度神经网络/滞回曲线/骨架曲线Key words
fiber reinforced materials/reinforced concrete shear wall/seismic performance/multivariate time series deep neural network/hysteresis curve/backbone curve分类
建筑与水利引用本文复制引用
贺畅,孔庆钊,熊青松,袁程..基于生成对抗网络的FRP加固钢筋混凝土剪力墙滞回性能预测方法[J].湖南大学学报（自然科学版）,2026,53(7):89-103,15.基金项目
国家自然科学基金资助项目(52108470),National Natural Science Foundation of China(52108470) （52108470）
国家留学基金项目(202406260146),China Scholarship Council Program(202406260146) （202406260146）
