Identifying method of rigid pavement support deterioration based on local CPSD-XGBoost fusion
-
摘要: 针对机场刚性道面板底支承劣化识别中传统动力学指标鲁棒性不足及纯数据驱动模型物理可解释性差、易过拟合的问题，提出了一种数据-物理融合的识别方法。通过理论推导明确了互功率谱密度（简称互谱）与道面支承状态的相关性，并面向分布式振动感知提出了区域互谱；区域互谱与XGBoost结合建立了数据物理融合方法，分步识别支承劣化板块及其具体位置；通过对比物理指标法、PCA-SVM数据驱动法及本融合方法在2个试验场景下的表现，验证了其有效性。分析结果表明：区域互谱综合反映了结构振型、频率、阻尼比对道面支承状况的影响，且加载点对支承异常的敏感性显著高于互谱计算点；在识别性能上，物理指标法的平均准确率（83.35%）和召回率（50%）最低；数据驱动方法虽准确率达90.75%，但实际数据可分性较弱且特征冗余（7~14个）；数据-物理融合方法表现最优，准确率提升至91.65%，召回率达77.5%，特征维度大幅降至3~5个，特征对应于区域互谱对支承状态的敏感频段，物理可解释性强。研究为刚性道面支承劣化识别提供了一种将物理信息与数据特征结合的方法，在改进既有物理指标的同时，也为人工智能方法在道面结构异常识别的应用提供了可能的途径。Abstract: In response to insufficient robustness in traditional dynamic indicators and poor physical interpretability or over-fitting in purely data-driven models for identifying the deterioration of airport rigid pavement slab support, a data-physics fusion identification method was proposed. The correlation between cross power spectral density (CPSD) and the pavement support state was clarified through theoretical derivation. For distributed vibration sensing, local CPSD was proposed. A data-physics fusion method was developed by combining local CPSD with XGBoost to identify slabs with deteriorated support and subsequently locate the specific regions. Its effectiveness was validated by comparing the performance of the physical indicator method, the principal component analysis-support vector machine (PCA-SVM) data-driven method, and the proposed fusion method across two experimental scenarios. Analysis results show that local CPSD comprehensively reflects the influence of structural mode shapes, frequencies, and damping ratios on the pavement support state. Furthermore, the sensitivity of loading points to support anomalies is significantly higher than that of the CPSD calculation points. Regarding identification performance, the physical indicator method has the lowest average accuracy (83.35%) and recall (50%). Although the data-driven method achieves an accuracy of 90.75%, the actual data exhibit weak separability and feature redundancy (7-14 features). The data-physics fusion method performs the best, with the accuracy rising to 91.65%, the recall reaching 77.5%, and the feature dimension significantly reducing to 3-5. These features correspond to the sensitive frequency bands of the local CPSD for support states, providing strong physical interpretability. A method is provided to combine physical information with data features for identifying support deterioration in rigid pavements. While improving existing physical indicators, it also offers a possible path for applying artificial intelligence methods to identify anomalies in pavement structures.
-
图 1 板底支承状态对互谱的影响
Figure 1. Influence of support conditions of slab on cross power spectral density
图 2 道面支承劣化识别的顺序与方法对比
Figure 2. Order and method comparison of pavement support deterioration identification
图 6 脱空区的计算点与加载点对区域互谱的影响分析
Figure 6. Influence analysis of calculation points and loading points in void area on local CPSD
图 9 前2个主成分的二维可视化与板的识别结果
Figure 9. 2D visualization of the first two principal components and slab identification
图 10 XGBoost特征排序中不同频率点的贡献度
Figure 10. Contribution of different frequency points in XGBoost feature ordering
图 11 单板足尺试验中各样本的频带能量占比及脱空位置识别结果
Figure 11. Band energy ratio of samples and identification results of void position in one-slab full-scale tests
图 12 前2个主成分的二维可视化与单元的识别结果
Figure 12. 2D visualization of the first two principal components and void unit identification
图 13 XGBoost特征排序中各频率点的累积增益
Figure 13. Cumulative gains of frequency points in XGBoost feature ordering
图 15 数据物理融合方法的脱空位置识别结果
Figure 15. Void position identification results of data-physics fusion method
表 1 不同方法对支承劣化板的识别效果对比
Table 1. Comparison of identification performances of methods on slabs with deteriorated support
方法 预测结果 物理指标构建 PCA+SVM PCA+XGBoost XGBoost排序分类 健康 异常 健康 异常 健康 异常 健康 异常 实际结果 健康 24 0 22 2 22 2 22 2 异常 8 4 3 9 3 9 2 10 准确率/% 77.8 86.1 86.1 88.9 召回率/% 33.3 75.0 75.0 83.3 F1分数/% 50.0 78.3 78.3 83.3 
下载: 导出CSV
表 2 不同方法对支承劣化位置的识别效果对比
Table 2. Comparison of identification performance of methods on support deterioration position
方法 预测结果 物理指标构建 PCA+SVM PCA+XGBoost XGBoost排序分类 健康 异常 健康 异常 健康 异常 健康 异常 实际结果 健康 88 8 95 1 94 2 94 2 异常 4 8 4 8 4 8 4 8 准确率/% 88.9 95.4 94.4 94.4 召回率/% 66.7 66.7 66.7 66.7 F1分数/% 57.1 76.2 72.7 72.7
下载: 导出CSV
-
[1] 李盛, 张海涛, 孙煜, 等. 在役水泥路面劣化行为与延寿技术综述[J]. 交通运输工程学报, 2024, 24(3): 25-47. doi: 10.19818/j.cnki.1671-1637.2024.03.002LI Sheng, ZHANG Hai-tao, SUN Yu, et al. Review on deterioration behavior and life extension technologies of cement pavement in service[J]. Journal of Traffic and Transportation Engineering, 2024, 24(3): 25-47. doi: 10.19818/j.cnki.1671-1637.2024.03.002 [2] 凌建明, 刘海伦, 马正好, 等. 总弯沉比及其在机场刚性道面板底脱空判定中的适用性[J]. 同济大学学报(自然科学版), 2023, 51(7): 1085-1093.LING Jian-ming, LIU Hai-lun, MA Zheng-hao, et al. Total deflection ratio and its applicability in void identification of airport rigid pavement[J]. Journal of Tongji University (Natural Science), 2023, 51(7): 1085-1093. [3] ZHANG Y M, TONG Z, SHE X H, et al. SWC-net and multi-phase heterogeneous FDTD model for void detection underneath airport pavement slab[J]. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(12): 20698-20714. doi: 10.1109/TITS.2024.3459004 [4] 张一鸣, 鲍帆, 童峥, 等. 非均质机场水泥道面板底脱空雷达波响应[J]. 东南大学学报(自然科学版), 2023, 53(1): 137-148.ZHANG Yi-ming, BAO Fan, TONG Zheng, et al. Radar wave response of slab bottom voids in heterogeneous airport concrete pavement[J]. Journal of Southeast University(Natural Science Edition), 2023, 53(1): 137-148. [5] 张宇辉, 张献民. 机场道面及其下部地基脱空的测试方法[J]. 交通运输工程学报, 2016, 16(6): 1-11.ZHANG Yu-hui, ZHANG Xian-min. Test methods of airport pavement and subjacent foundation void[J]. Journal of Traffic and Transportation Engineering, 2016, 16(6): 1-11. [6] LI M X, ANDERSON N, SNEED L, et al. Condition assessment of concrete pavements using both ground penetrating radar and stress-wave based techniques[J]. Journal of Applied Geophysics, 2016, 135: 297-308. doi: 10.1016/j.jappgeo.2016.10.022 [7] WU D F, ZENG M Y, ZHAO H D, et al. A novel distributed sensing method for support condition monitoring under concrete pavement[J]. International Journal of Pavement Engineering, 2022, 23(7): 2227-2241. doi: 10.1080/10298436.2020.1849688 [8] 赵坪锐, 徐天赐, 刘卫星, 等. 单元板式轨道脱空伤损识别的柔度曲率特征值法[J]. 西南交通大学学报, 2021, 56(5): 1100-1108, 1127, 1109-1111.ZHAO Ping-rui, XU Tian-ci, LIU Wei-xing, et al. Flexibility curvature eigenvalue method for debonding damage identification of unit slab track[J]. Journal of Southwest Jiaotong University, 2021, 56(5): 1100-1108, 1127, 1109-1111. [9] ZENG M Y, WU D F, ZHAO H D, et al. Novel assessment method for support conditions of concrete pavement under traffic loads using distributed optical sensing technology[J]. Transportation Research Record, 2020, 2674(4): 42-56. doi: 10.1177/0361198120912994 [10] 曾孟源, 赵鸿铎, 吴荻非, 等. 基于振动感知的混凝土铺面板底脱空识别方法[J]. 中国公路学报, 2020, 33(3): 42-52.ZENG Meng-yuan, ZHAO Hong-duo, WU Di-fei, et al. Identification of cavities underneath concrete pavement based on pavement vibration[J]. China Journal of Highway and Transport, 2020, 33(3): 42-52. [11] 曾孟源, 赵鸿铎, 边泽英, 等. 基于分布式光纤的混凝土路面振动场感知与解析[J]. 中国公路学报, 2022, 35(7): 78-90.ZENG Meng-yuan, ZHAO Hong-duo, BIAN Ze-ying, et al. Sensing and analysis of concrete pavement vibration field based on distributed optical fiber[J]. China Journal of Highway and Transport, 2022, 35(7): 78-90. [12] ZHAO H D, WU D F, ZENG M Y, et al. Assessment of concrete pavement support conditions using distributed optical vibration sensing fiber and a neural network[J]. Construction and Building Materials, 2019, 216: 214-226. doi: 10.1016/j.conbuildmat.2019.04.195 [13] 吴荻非, 向晖, 刘成龙, 等. 基于振动传递率函数的水泥混凝土铺面脱空识别方法[J]. 北京工业大学学报, 2024, 50(4): 453-465.WU Di-fei, XIANG Hui, LIU Cheng-long, et al. Void-underneath identification method of cement concrete pavement using vibration transmissibility function[J]. Journal of Beijing University of Technology, 2024, 50(4): 453-465. [14] 罗丹, 黄晓琴, 冷费贤, 等. 数字孪生在交通基础设施智能建造中的应用与挑战[J]. 交通运输工程学报, 2025, 25(3): 33-64. doi: 10.19818/j.cnki.1671-1637.2025.03.003LUO Dan, HUANG Xiao-qin, LENG Fei-xian, et al. Applications and challenges of digital twin in intelligent construction of transportation infrastructure[J]. Journal of Traffic and Transportation Engineering, 2025, 25(3): 33-64. doi: 10.19818/j.cnki.1671-1637.2025.03.003 [15] SONODA Y, LU C, YIN Y F. Basic research on usefulness of convolutional autoencoders in detecting defects in concrete using hammering sound[J]. Structural Health Monitoring, 2023, 22(4): 2231-2250. doi: 10.1177/14759217221122296 [16] 张军, 姜文涛, 张云, 等. 基于极限梯度提升和探地雷达时频特征的水泥路面脱空识别[J]. 同济大学学报(自然科学版), 2024, 52(1): 104-114, 121.ZHANG Jun, JIANG Wen-tao, ZHANG Yun, et al. Cement pavement void identification based on XGBoost and gpr time-frequency features[J]. Journal of Tongji University (Natural Science), 2024, 52(1): 104-114, 121. [17] ZHANG J, LU Y M, YANG Z, et al. Recognition of void defects in airport runways using ground-penetrating radar and shallow CNN[J]. Automation in Construction, 2022, 138: 104260. doi: 10.1016/j.autcon.2022.104260 [18] LI H F, WANG B Y, LIU S S, et al. GPR-STA: A style transfer algorithm for enhancing GPR data in airport runway structural defect detection[C]//Springer. CCF National Conference of Computer Applications. Singapore: Springer, 2024: 344-358. [19] SHI B, WANG X, DONG Q, et al. Voids prediction beneath cement concrete slabs using a FEM-ANN method[J]. International Journal of Pavement Engineering, 2023, 24(1): 2191198. doi: 10.1080/10298436.2023.2191198 [20] ZHU Y J, CHEN L Y. Framework for developing prediction models for boundary conditions of slabs in girders considering interpretability: An application for deck slabs in composite box girders[J]. Journal of Bridge Engineering, 2025, 30(4): 04025010. doi: 10.1061/JBENF2.BEENG-6935 [21] IJAZ H, QABUR A, AATI K, et al. Physics-aware mesh-free deep learning for rectangular elastic plate deformation with support-dependent boundary enforcement[J]. International Journal of Applied Mechanics, 2025, 17(12): 2550109. doi: 10.1142/S1758825125501091 [22] ZHOU W, XU Y F. Damage identification for plate structures using physics-informed neural networks[J]. Mechanical Systems and Signal Processing, 2024, 209: 111111. doi: 10.1016/j.ymssp.2024.111111 [23] 余忠儒, 单德山, 孙榕徽. 基于种群的桥梁结构健康监测研究综述与挑战[J]. 交通运输工程学报, 2025, 25(5): 1-22. doi: 10.19818/j.cnki.1671-1637.2025.05.001YU Zhong-ru, SHAN De-shan, SUN Rong-hui. Population-based structural health monitoring of bridges: Review and challenges [J]. Journal of Traffic and Transportation Engineering, 2025, 25(5): 1-22. doi: 10.19818/j.cnki.1671-1637.2025.05.001 [24] SAJID S, CHOUINARD L, CARINO N. Condition assessment of concrete plates using impulse-response test with affinity propagation and homoscedasticity[J]. Mechanical Systems and Signal Processing, 2022, 178: 109289. doi: 10.1016/j.ymssp.2022.109289 [25] WANG X, MA Z L, HU X, et al. Void detection of airport concrete pavement slabs based on vibration response under moving load[J]. Sensors, 2025, 25(15): 4703. doi: 10.3390/s25154703 [26] LIU X Z, ZHU L Y, LI Z W. Dynamic assessment of interlayer debonding in high-speed slab tracks using frequency-domain simulation[J]. Results in Engineering, 2026, 29: 108515. doi: 10.1016/j.rineng.2025.108515 [27] TANG L Q, LI Y H, BAO Q, et al. Quantitative identification of damage in composite structures using sparse sensor arrays and multi-domain-feature fusion of guided waves[J]. Measurement, 2023, 208: 112482. doi: 10.1016/j.measurement.2023.112482 [28] LIU J, LIU F Y, WANG L B. Automated, economical, and environmentally-friendly asphalt mix design based on machine learning and multi-objective grey wolf optimization[J]. Journal of Traffic and Transportation Engineering (English Edition), 2024, 11(3): 381-405. doi: 10.1016/j.jtte.2023.10.002 [29] LAXMAN K C, TABASSUM N, AI L, et al. Automated crack detection and crack depth prediction for reinforced concrete structures using deep learning[J]. Construction and Building Materials, 2023, 370: 130709. doi: 10.1016/j.conbuildmat.2023.130709 [30] BABOLI NEZHADI E, LABIBZADEH M, HOSSEINLOU F, et al. Machine learning-based design of double corrugated steel plate shear walls[J]. International Journal of Structural Integrity, 2024, 15(6): 1216-1248. doi: 10.1108/IJSI-09-2024-0152 [31] YUAN H Z, ZHAO K, YAN Y, et al. Long tunnel group driving fatigue detection model based on XGBoost algorithm[J]. Journal of Traffic and Transportation Engineering (English Edition), 2025, 12(1): 167-179. doi: 10.1016/j.jtte.2023.02.008 -
计量
- 文章访问数: 15
- HTML全文浏览量: 7
- PDF下载量: 1
- 被引次数: 0
