横向撞击下钢管混凝土构件峰值位移的机器学习预测及可解释性分析

钟学琦; 陈雨凡; 雷宇龙; 廖飞宇

doi:10.13206/j.gjgS26042702

横向撞击下钢管混凝土构件峰值位移的机器学习预测及可解释性分析

doi: 10.13206/j.gjgS26042702

福建农林大学交通与土木工程学院, 福州 350108

基金项目:

国家自然科学基金项目（52378140）；福建省青年科技人员育成项目（KJBX25090A）；福建省自然科学基金创青项目（2026J008106）。

详细信息

作者简介:
钟学琦，博士，副教授，主要研究方向为桥梁工程。

通讯作者:
廖飞宇，博士，教授，主要研究方向为组合结构，feiyu.liao@fafu.edu.cn。

计量
- 文章访问数: 2
- HTML全文浏览量: 0
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2026-04-27
- 网络出版日期: 2026-07-06

Machine Learning Prediction and Interpretability Analysis of Peak Displacement of Concrete-Filled Steel Tube Members Under Lateral Impact

College of Transportation and Civil Engineering, Fujian Agriculture and Forest University, Fuzhou 350108, China

摘要

摘要: 针对钢管混凝土（CFST）构件在横向撞击作用下峰值位移难以快速、准确评估，以及现有数据驱动方法对模型内部决策机制解释不足的问题，基于国内外横向撞击试验文献建立了CFST构件峰值位移数据库，并构建了兼具预测能力与可解释性的机器学习模型。数据库共包含193组有效试验样本，涵盖构件几何参数、材料性能参数和撞击工况参数，选取落锤质量、冲击速度、截面直径、钢管厚度、轴压比、钢管屈服强度、混凝土抗压强度和试件长度8个参数作为输入特征，以跨中峰值位移作为输出特征。采用6种集成学习算法建立峰值位移预测模型，并利用Optuna框架结合五折交叉验证进行超参数自动寻优。同时，为提高模型结果的透明性，采用沙普利加和解释（Shapley Additive Explanations，SHAP）方法对CatBoost模型进行解释性分析，以量化各输入参数对峰值位移预测结果的贡献程度、影响方向及变量间的潜在交互关系。结果表明：在6种集成学习模型中，CatBoost模型综合表现最优，能够较好表征多参数耦合作用下CFST构件峰值位移的非线性变化规律；冲击速度对峰值位移预测结果的贡献最大，是主导参数；落锤质量和截面直径的重要性次之，轴压比影响相对较弱；冲击速度、落锤质量和试件长度整体上促进峰值位移增大，而截面直径、钢管厚度和钢管屈服强度整体上抑制峰值位移发展；混凝土抗压强度和轴压比在当前数据范围内影响相对有限；钢管屈服强度与混凝土抗压强度之间存在较为显著的交互作用。
- 钢管混凝土构件 /
- 横向撞击 /
- 峰值位移 /
- 机器学习 /
- SHAP可解释性分析
Abstract: To address the difficulty in rapidly and accurately evaluating the peak displacement of concrete-filled steel tube （CFST） members under lateral impact， as well as the insufficient interpretation of the internal decision-making mechanism of existing data-driven methods， a peak-displacement database of CFST members was established based on published domestic and international lateral impact test data， and a machine learning model that balances predictive accuracy and interpretability was developed. The database consists of 193 valid test samples， covering geometric parameters， material properties， and impact conditions. Eight parameters， namely hammer mass， impact velocity， section diameter， steel tube thickness， axial compression ratio， steel yield strength， concrete compressive strength， and member length—were selected as input features， and the mid-span peak displacement was taken as the output feature. Based on the experimental database， six ensemble learning algorithms were employed to establish peak-displacement prediction models， and the Optuna framework combined with five-fold cross-validation was used for automatic hyperparameter optimization. Meanwhile， To enhance the transparency of the model results， Shapley Additive Explanations （SHAP） were employed to conduct an interpretability analysis of the CatBoost model， quantifying the contribution degree， influence direction， and potential interactions among variables of each input parameter on the predicted peak displacement. The results indicate that among the six ensemble learning models， the CatBoost model exhibits the best overall performance， effectively characterizing the nonlinear variation pattern of peak displacement in CFST members under multi-parameter coupling. Impact velocity contributes the most to the prediction of peak displacement and serves as the dominant parameter. Hammer mass and section diameter rank next in importance， whereas the influence of the axial compression ratio is relatively weak. Overall， impact velocity， hammer mass， and specimen length tend to amplify the peak displacement， while section diameter， steel tube thickness， and steel yield strength generally mitigate its development. The effects of concrete compressive strength and axial compression ratio remain relatively limited within the current data range. Furthermore， a significant interaction exists between steel yield strength and concrete compressive strength.
- concrete-filled steel tube members /
- lateral impact /
- peak displacement /
- machine learning /
- SHAP interpretability analysis

HTML全文

参考文献(42)

[1]	张玉芬, 郝圣旺, 李栋磊, 等. 复式钢管混凝土墩柱轴压性能及承载力提高分析[J]. 工程力学, 2025, 42（1）:209-222.
[2]	Han L H, Li W, Bjorhovde R. Developments and advanced applications of concrete-filled steel tubular（CFST）structures:members[J]. Journal of Constructional Steel Research, 2014, 100:211-228.
[3]	Wang R, Han L H, Hou C C. Behavior of concrete filled steel tubular（CFST）members under lateral impact:experiment and FEA model[J]. Journal of Constructional Steel Research, 2013, 80:188-201.
[4]	梅圣琪, 王蕊, 赵晖, 等. 撞击作用下圆端形钢管混凝土动力响应与挠度预测[J]. 工程力学, 2025, 42（9）:51-62.
[5]	刘奇, 江俐敏, 罗浩. 桥梁结构受货船撞击后的影响分析[J]. 山西建筑, 2022, 48（18）:145-148.
[6]	National Transportation Safety Board. Contact of containership dali with the francis scott key bridge and subsequent bridge collapse[R]. Washington D C:National Transportation Safety Board, 2024:1-50.
[7]	Bambach M R. The behaviour of hollow and concrete filled thin-walled steel tubular members subjected to transverse impact[J]. Thin-Walled Structures, 2011, 49（10）:1259-1268.
[8]	姜珊, 路国运, 杨会伟. 侧向冲击载荷下钢管混凝土结构的动力响应及参数分析[J]. 爆炸与冲击, 2023, 43（11）:28-39.
[9]	Roy D, Das D, Islam K, et al. Machine learning-based prediction of CFST columns using gradient tree boosting algorithm[J]. Composite Structures, 2021, 259:113505.
[10]	Lai D, Demartino C, Xiao Y. Probabilistic machine leaning models for predicting the maximum displacements of concrete-filled steel tubular columns subjected to lateral impact loading[J]. Engineering Applications of Artificial Intelligence, 2024, 135:108704.
[11]	Lai D, Wei J, Contento A, et al. Machine learning-based probabilistic predictions for concrete filled steel tube（CFST）column axial capacity[J]. Structures, 2024, 70:107543.
[12]	Afshar A, Nouri G, Ghazvineh S, et al. Machine-learning applications in structural response prediction:a review[J]. Practice Periodical on Structural Design and Construction, 2024, 29（3）:1-15.
[13]	Tiwari K, Kumar R, Singh P, et al. SHAP for shear strength predictions in deep RC beams[J]. Engineering Structures, 2024, 300:117000.
[14]	Shang Y, Wang Y, Li M, et al. Explainable AI integration into hybrid ML frameworks for confined columns[J]. Structures, 2025, 75:108000.
[15]	Elhishi S, Elashry A M, El-Metwally S. Unboxing machine learning models for concrete strength prediction using XAI[J]. Scientific Reports, 2023, 13（1）:19892.
[16]	徐继伟, 杨云. 集成学习方法:研究综述[J]. 云南大学学报（自然科学版）, 2018, 40（6）:1082-1092.
[17]	雷相东. 机器学习算法在森林生长收获预估中的应用[J]. 北京林业大学学报, 2019, 41（12）:23-36.
[18]	郭衍昊, 窦杰, 向子林, 等. 基于优化负样本采样策略的梯度提升决策树与随机森林的汶川同震滑坡易发性评价[J]. 地质科技通报, 2024, 43（3）:251-265.
[19]	温泉, 余玉欢, 庄尚德, 等. 融合SHAP和TSO-XGBoost模型的水路货运量预测[J]. 水利水运工程学报, 2024（6）:86-96.
[20]	李珊珊, 孙朝阳, 李国栋. 基于LightGBM与SHAP的空腔积水深度可解释性机器学习模型[J]. 力学季刊, 2024, 45（2）:442-453.
[21]	王庆志, 张秀华, 吴吉展, 等. 基于CatBoost算法的渗碳齿轮接触疲劳极限预测方法研究[J]. 力学学报, 2024, 56（12）:3564-3576.
[22]	Sheng H, Ren Z, Wang D, et al. Estimation and interpretation of interfacial bond in concrete-filled steel tube by using optimized XGBoost and SHAP[J]. Structures, 2024, 70:107669.
[23]	Du G, Babic M, Wu F, et al. Experimental and numerical studies on concrete filled circular steel tubular（CFCST）members under impact loads[J]. International Journal of Civil Engineering, 2019, 17（8）:1211-1226.
[24]	Bambach M R, Jama H, Zhao X L, et al. Hollow and concrete filled steel hollow sections under transverse impact loads[J]. Engineering Structures, 2008, 30（10）:2859-2870.
[25]	Yousuf M, Uy B, Tao Z, et al. Impact behaviour of pre-compressed hollow and concrete filled mild and stainless steel columns[J]. Journal of Constructional Steel Research, 2014, 96:54-68.
[26]	Shakir A S, Guan Z W, Jones S W. Lateral impact response of the concrete filled steel tube columns with and without CFRP strengthening[J]. Engineering Structures, 2016, 116:148-162.
[27]	任够平, 李珠, 王蕊. 低速侧向冲击下钢管混凝土柱挠度研究[J]. 工程力学, 2008, 25（5）:170-175.
[28]	王蕊, 李珠, 任够平, 等. 钢管混凝土梁在侧向冲击荷载作用下动力响应的试验研究和数值模拟[J]. 土木工程学报, 2007, 40（10）:34-40.
[29]	李珠, 王瑞峰. 固简支钢管混凝土构件侧向冲击试验研究[J]. 工程力学, 2008, 25（增刊1）:193-197.
[30]	Han L H, Hou C C, Zhao X L, et al. Behaviour of high-strength concrete filled steel tubes under transverse impact loading[J]. Journal of Constructional Steel Research, 2014, 92:25-39.
[31]	Yang X, Yang H, Zhang S. Transverse impact behavior of high-strength concrete filled normal-/high-strength square steel tube columns[J]. International Journal of Impact Engineering, 2020, 139:103512.
[32]	Wang W, Wu C, Li J, et al. Behavior of ultra-high performance fiber-reinforced concrete（UHPFRC）filled steel tubular members under lateral impact loading[J]. International Journal of Impact Engineering, 2019, 132:103314.
[33]	Yang Y F, Zhang Z C, Fu F. Experimental and numerical study on square RACFST members under lateral impact loading[J]. Journal of Constructional Steel Research, 2015, 111:43-56.
[34]	Mi Y, Liu Z, Wang W, et al. Experimental study on residual axial bearing capacity of UHPFRC-filled steel tubes after lateral impact loading[J]. Structures, 2020, 27:549-561.
[35]	Deng Y, Tuan C Y, Xiao Y. Flexural behavior of concrete-filled circular steel tubes under high-strain rate impact loading[J]. Journal of Structural Engineering, 2012, 138（3）:449-456.
[36]	Wang Y, Qian X, Liew J Y R, et al. Impact of cement composite filled steel tubes:an experimental, numerical and theoretical treatise[J]. Thin-Walled Structures, 2015, 87:76-88.
[37]	Merwad A M, El-Sisi A A, Mustafa S A A, et al. Lateral impact response of rubberized-fibrous concrete-filled steel tubular columns:experiment and numerical study[J]. Buildings, 2022, 12（10）:1566.
[38]	Alam M I, Fawzia S, Zhao X L, et al. Performance and dynamic behaviour of FRP strengthened CFST members subjected to lateral impact[J]. Engineering Structures, 2017, 147:160-176.
[39]	Remennikov A M, Kong S Y, Uy B. Response of foam-and concrete-filled square steel tubes under low-velocity impact loading[J]. Journal of Performance of Constructed Facilities, 2011, 25（5）:373-381.
[40]	Wu H, Ren G M, Fang Q, et al. Response of ultra-high performance cementitious composites filled steel tube（UHPC-CFST）subjected to low-velocity impact[J]. International Journal of Impact Engineering, 2019, 132:103314.
[41]	王艺凡, 阳霞. 螺旋筋增强耐候钢管混凝土柱抗侧向冲击性能试验研究[J]. 建筑结构, 2023, 53（2）:76-84.
[42]	Badawy A, Eltobgy H H, Darwish E, et al. Impact response of concrete-filled steel tubular members（CFST）using different types of concrete filling[J]. International Journal of Scientific& Technology Research, 2021, 10（6）:288-303.