基于BP神经网络的FRP筋混凝土板抗冲切承载力预测模型  

The punching shear capacity prediction model for FRP reinforced concrete slabs based on BP neural network

作　　者：沈袁协 梁诗雪 SHEN Yuanxie;LIANG Shixue(School of Civil Engineering and Architecture,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学建筑工程学院,杭州310018

出　　处：《浙江理工大学学报（自然科学版）》2022年第3期441-451,共11页Journal of Zhejiang Sci-Tech University(Natural Sciences)

基　　金：浙江省自然科学基金项目(LY22E080016,LGF20E080019)。

摘　　要：为了改善现有FRP筋混凝土板抗冲切承载力预测模型因影响因素考虑不全而导致泛化性能较差的问题,建立了以数据驱动为核心的FRP筋混凝土板抗冲切承载力预测模型。首先收集了121组FRP筋混凝土板柱节点抗冲切承载力数据,采用BP神经网络建立了FRP筋混凝土板抗冲切承载力预测模型;然后采用Garson算法对影响FRP筋混凝土板抗冲切承载力的因素进行敏感性分析。将该模型与其他传统承载力计算公式的预测结果进行对比,结果表明该模型的预测结果最好,误差更小。与英国规范相比,均方根误差降低了29.7%,平均绝对百分比误差降低了21.5%,判定系数提升了3.6%。敏感性分析的结果验证了输入参数选取的合理性并得出了各参数的影响性排序,发现板的有效高度对冲切承载力的影响最为显著。该研究可为FRP筋板柱节点抗冲切性能的分析模型和精细化设计提供帮助。In order to improve the poor generalization performance of punching shear capacity prediction model for FRP reinforced concrete slabs due to inadequate consideration of the influencing factors, a data drive-based model for predicting the punching shear capacity of FRP reinforced concrete slabs was established. Firstly, 121 groups of punching shear capacity data on FRP reinforced concrete slab-column joints were collected, and a punching shear capacity prediction model for FRP reinforced concrete slabs based on BP neural network was established. Secondly, the sensitivity analysis was performed on factors affecting the punching shear capacity of FRP reinforced concrete slabs using Garson algorithm. Through comparison with the prediction results of other traditional bearing capacity computing formulas, it was found that the model proposed in this paper had the optimal prediction results, with less errors. Compared with the British specification, the root-mean-square error reduced by 29.7%, mean absolute percentage error decreased by 21.5%, and coefficient of determination improved by 3.6%. The sensitivity analysis results have verified the rational selection of the input parameters and obtained the sort of the influence degree of each parameter. This study has found that the effective height of the concrete slab has the most significant impact on the punching shear capacity of FRP reinforced concrete slabs. The study is expected to support the analysis model and fine design of the punching shear performance of FRP reinforced concrete slab-column joints.

关 键 词：BP神经网络 FRP筋混凝土板 机器学习 抗冲切承载力 Garson算法 

分 类 号：TU377.9[建筑科学—结构工程]