A prediction model for bonding strength of asphalt overlay on airport pavement based on texture feature data
-
摘要: 为精准预测“白改黑”机场道面层间黏结强度并揭示界面纹理特征与其的关联机制，本研究制备了16种界面处治工况的旧混凝土面板，通过三维激光检测技术提取界面的几何、波谱和分形指标，建立涵盖几何幅值-尺度分布-结构复杂度的纹理多元表征指标体系；加铺沥青混合料形成“白改黑”复合试件后在3种试验温度下进行剪切试验，构建基于多元纹理数据驱动的剪切强度标准化数据集；基于多层感知机(MLP)和极限梯度提升(XGBoost)建立了集成机器学习模型，并提出改进浣熊优化算法(ICOA)实现超参数寻优；最后利用SHAP分析量化各特征参数对预测结果的贡献。研究结果表明：铣刨深度为6~10 mm时，在50 ℃条件下强度仍可维持在0.65 MPa，相比于铣刨深度为4~6 mm时提升了67.9%~69.5%，可显著提升机场高温抗剪能力；ICOA-MLP-XGBoost模型相比其他模型，其决定系数提升了6.1%~10.9%，平均绝对误差降低了35.8%~46.9%，表现出更优越的预测精度和泛化能力，且具有更好的泛化性能和工程稳定性；SHAP分析表明，平均纹理深度、谱熵和分形维数是影响剪切强度的主要因素，鉴于机场道面高胎压重载、低速工况下剪应力集中，建议以6.0~8.0 mm铣刨深度对界面进行处治以达到这3个指标的高SHAP贡献区域。本研究为“白改黑”机场道面层间性能预测提供了高精度、可解释的数据驱动方法，可指导界面处治方案优化与道面耐久性提升。Abstract: To accurately predict the bonding strength of "white-to-black" airport pavement interfaces and reveal the underlying mechanism linking interface texture to bonding performance, aged concrete slabs under 16 different treatment conditions were prepared. Three-dimensional laser scanning was used to extract geometric, spectral, and fractal indicators, establishing a multi-index texture characterization system covering geometric amplitude, distribution, and structural complexity. After placing asphalt mixture overlay to form "white-to-black" composite specimens, shear tests were conducted at three temperatures. A standardized shear strength dataset driven by multi-dimensional texture data was constructed. An ensemble machine-learning model combining a multilayer perceptron (MLP) and extreme gradient boosting (XGBoost) was developed, and an improved raccoon optimization algorithm (ICOA) was proposed for hyperparameter tuning. Finally, Shapley Additive Explanations (SHAP) was applied to quantify the contribution of each feature parameter to the predictions. The results show that when the milling depth is 6-10 mm, the shear strength can still maintain 0.65 MPa at 50 ℃. Compared with a milling depth of 4-6 mm, the strength increases by 67.9%-69.5%, indicating a significant improvement in high-temperature shear resistance for airport pavements. The ICOA-MLP-XGBoost model outperforms other models, with R2 improved by 6.1%-10.9% and MAE reduced by 35.8%-46.9%. It achieved higher prediction accuracy and better generalization, with improved robustness for engineering use. SHAP analysis indicates that mean texture depth, spectral entropy, and fractal dimension are the dominant factors affecting shear strength. Given the shear-stress concentration under heavy loads with high tire pressure at low speed on airport pavements, an interface milling depth of 6.0-8.0 mm is recommended for interface treatment to achieve high SHAP contribution ranges for these three indicators. This study provides a highly accurate and interpretable data-driven method for predicting interface performance in "white-to-black" airport pavements, which can guide interface-treatment optimization and improve pavement durability.
-
图 1 旧水泥混凝土板界面铣刨处治
Figure 1. Milling treatment of the interface of old cement concrete slabs
图 2 抛丸前后水泥板界面对比
Figure 2. Comparison of the cement board interface before and after shot blasting
图 4 界面高程点云数据提取与预处理
Figure 4. Extraction and preprocessing of interface elevation point cloud data
图 11 抛丸处治后水泥板界面纹理特征指标数值
Figure 11. Value of the texture characteristic index of the cement board interface after shot blasting treatment
图 12 铣刨处治后水泥板界面纹理特征指标数值
Figure 12. Value of the interface texture characteristic index of the cement board after milling treatment
图 13 不同层间界面处治工况下“白改黑”道面的剪切强度
Figure 13. Shear strength of the white-to-black runway under different interlayer interface treatment conditions
图 14 MLP-XGBoost模型经过超参数优化后的性能对比
Figure 14. Performance comparison of the MLP-XGBoost model after hyperparameter optimization
图 15 ICOA-MLP-XGBoost模型在测试集上的预测结果与实测结果对比
Figure 15. Comparison of the prediction results of the ICOA-MLP-XGBoost model on the test set with the measured results
图 16 MLP-XGBoost模型经过不同超参数算法优化后5折交叉验证结果
Figure 16. 5-fold cross-validation results of the MLP-XGBoost model after being optimized with different hyperparameter algorithms
表 1 BE-2型乳化沥青技术指标
Table 1. Technical specifications of BE-2 emulsified asphalt
类型 筛上剩余量/% 破乳速度 标准黏度/s 蒸发残留物性质 固体质量含量/% 25 ℃针入度/0.1 mm 软化点/℃ 10 ℃延度/cm BE-2 ＜0.1 中裂 26.5 56.87 72.8 46.5 >100 
下载: 导出CSV
表 2 水泥配合比设计
Table 2. Cement mix proportion design
指标 材料用量/(kg·m-3) 砂率/% 水灰比 水泥 砂 碎石 水 粒径4.75~9.5 mm 粒径9.5~19 mm 数值 310 650 530 800 150 33 0.48
下载: 导出CSV
表 3 抛丸处治参数
Table 3. Shot blasting treatment parameters
编号 抛丸次数 平均纹理深度/mm 编号 抛丸次数 平均纹理深度/mm YZ 0 0.11 P-6 6 0.30 P-1 1 0.36 P-7 7 0.29 P-2 2 0.33 P-8 8 0.30 P-3 3 0.29 P-9 9 0.28 P-4 4 0.27 P-10 10 0.29 P-5 5 0.28
下载: 导出CSV
表 4 铣刨处治参数
Table 4. Milling treatment parameters
mm 编号 铣刨深度 平均纹理深度 YZ 0 0.11 X-1 (0, 2] 0.51 X-2 (2, 4] 0.60 X-3 (4, 6] 0.82 X-4 (6, 8] 1.11 X-5 (8, 10] 1.19
下载: 导出CSV
表 5 沥青加铺层级配及沥青用量(SAM-13)
Table 5. Asphalt overlay gradation and asphalt dosage (SAM-13)
不同筛孔孔径(mm)下的质量百分率/% 沥青用量/% 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.075 100 96 56 29 22 18 15 12 10 6.2
下载: 导出CSV
表 6 输入和输出变量的统计分析
Table 6. Statistical analysis of input and output variables
指标 最大值 最小值 平均值 中位数 标准差 平均纹理深度/mm 1.327 0.102 0.463 0.328 0.309 算术平均高度/mm 1.070 0.130 0.430 0.336 0.238 偏度 6.921 1.222 4.531 5.129 1.640 中心频率/(次·mm-1) 1.349 0.486 0.911 0.877 0.218 谱熵 0.008 94 0.000 17 0.001 97 0.001 27 0.002 07 分形维数 2.133 1.202 1.750 1.759 0.223 试验温度/℃ 50 0 25 25 20 剪切强度/MPa 2.671 0.058 1.334 1.408 0.764
下载: 导出CSV
表 7 MLP结构参数设置
Table 7. Structure parameter settings of MLP
结构参数 搜索范围 num_hidden_layers [1, 5]，基线模型取值2 hidden_units [32, 128]，基线模型取值64 learning_rate [0.001, 0.5]，基线模型取值0.01 batch_size [16, 256]，基线模型取值32 dropout_rate [0.1, 0.5]，基线模型取值0.1 weight_decay [1×10-4, 1×10-2]，基线模型取值1×10-3
下载: 导出CSV
表 8 XGBoost结构参数设置
Table 8. Structure parameter settings of XGBoost
结构参数 搜索范围 n_estimators [50, 500]，基线模型取值2 max_depth [1, 12]，基线模型取值6 learning_rate [0.01, 0.2]，基线模型取值0.05 subsample [0.5, 1]，基线模型取值0.8 colsample_bytree [0.5, 1]，基线模型取值0.8 min_child_weight [1, 20]，基线模型取值2
下载: 导出CSV
表 9 机器学习算法
Table 9. Machine learning algorithm
序号 名称 优势 参考文献 1 支持向量回归(Support Vector Regression, SVR) 泛化强，非线性拟合好 [21] 2 随机森林回归(Random Forest, RF) 抗噪稳健强，不易过拟合 [45] 3 最小二乘提升(Least Squares Boosting, LSBoost) 精度提升明显，适应非线性 [46] 4 广义加性模型(Generalized Additive Model, GAM) 可刻画，非线性 [47] 5 极限学习机回归(Extreme Learning Machine Regression, ELM) 训练速度极快，实现简单高效 [26] 6 梯度提升树(Light Gradient Boosting Machine, LGBM) 训练高效快速，精度高可扩展 [48]
下载: 导出CSV
表 10 超参数优化算法
Table 10. Hyperparameter optimization algorithms
序号 名称 优势 参考文献 1 粒子群优化算法(Particle Swarm Optimization, PSO) 易于实现，快速收敛，良好的探索开发，可并行性 [50] 2 蚁群优化算法(Ant Colony Optimization, ACO) 强组合问题求解器，通过信息素反馈自适应，抗局部最优鲁棒，固有可并行化 [50] 3 白鲸优化算法(Whale Optimization Algorithm, WOA) 高效的搜索机制，快速的收敛速度，最小参数设置，优秀的探索-开发平衡 [50] 4 贝叶斯优化算法(Bayesian Optimization, BO) 高效的全局优化，数据高效的函数优化，平衡探索和开发 [51] 5 饥饿游戏优化算法(Hunger Games Search, HGS) 自适应搜索，全局局部平衡，计算简单，鲁棒性能优秀 [52] 6 海鹰搜索算法(Cald Eagle Search, BES) 有效的探索能力，稳健的全局优化性能，有效的探索-开发平衡 [52]
下载: 导出CSV
表 11 9种基线模型在特征集A测试集的预测性能对比
Table 11. Comparison of the prediction performance of test sets of feature set A of nine baseline models
模型 MAE MAPE RMSE R2 SVR 0.211 0.291 0.290 0.677 RF 0.215 0.296 0.299 0.662 LSBoost 0.206 0.283 0.282 0.683 GAM 0.222 0.309 0.307 0.652 ELM 0.230 0.315 0.319 0.633 LGBM 0.195 0.268 0.271 0.698 MLP 0.201 0.278 0.278 0.688 XGBoost 0.197 0.274 0.273 0.695 MLP-XGBoost 0.191 0.263 0.263 0.709
下载: 导出CSV
表 12 9种基线模型在特征集B测试集的预测性能对比
Table 12. Comparison of the prediction performance of test sets of feature set B of nine baseline models
模型 MAE MAPE RMSE R2 SVR 0.182 0.203 0.235 0.732 RF 0.188 0.231 0.247 0.715 LSBoost 0.179 0.195 0.222 0.754 GAM 0.195 0.266 0.268 0.704 ELM 0.207 0.276 0.281 0.678 LGBM 0.165 0.173 0.195 0.811 MLP 0.173 0.186 0.218 0.779 XGBoost 0.167 0.178 0.203 0.805 MLP-XGBoost 0.152 0.168 0.189 0.847
下载: 导出CSV
表 13 ICOA-MLP-XGBoost模型在2个特征集下的预测性能对比
Table 13. Comparison of the prediction performance of two feature sets of ICOA-MLP-XGBoost model
特征集 MAE MAPE RMSE R2 A 0.158 0.173 0.193 0.822 B 0.071 0.089 0.092 0.964
下载: 导出CSV
-
[1] 姜利, 苏禹, 程维, 等. 基于应力吸收层的"白改黑" 路面温度应力有限元分析[J]. 科学技术与工程, 2019, 19(3): 232-238.JIANG Li, SU Yu, CHENG Wei, et al. Finite element analysis of temperature stress of "white to black" pavement based on stress absorption layer[J]. Science Technology and Engineering, 2019, 19(3): 232-238. [2] 向泽君, 罗再谦, 汪明. 连续实景影像在"白改黑"工程测量上的应用研究[J]. 测绘通报, 2011(9): 42-44, 69.XIANG Ze-jun, LUO Zai-qian, WANG Ming. Application of digital panorama map to road project during the night work[J]. Bulletin of Surveying and Mapping, 2011(9): 42-44, 69. [3] WU C Y, HONG S X, CHEN T, et al. Application of hot-mixed ultrathin friction course in the white-to-black maintenance project of concrete bridge deck[J]. Journal of Performance of Constructed Facilities, 2022, 36: 04021116. doi: 10.1061/(ASCE)CF.1943-5509.0001697 [4] DAS R, MOHAMMAD L N, ELSEIFI M, et al. Effects of tack coat application on interface bond strength and short-term pavement performance[J]. Transportation Research Record: Journal of the Transportation Research Board, 2017, 2633(1): 1-8. [5] SALINAS A, AL-QADI I L, HASIBA K I, et al. Interface layer tack coat optimization[J]. Transportation Research Record: Journal of the Transportation Research Board, 2013(2372): 53-60. [6] RAPOSEIRAS A C, CASTRO-FRESNO D, VEGA-ZAMANILLO A, et al. Test methods and influential factors for analysis of bonding between bituminous pavement layers[J]. Construction and Building Materials, 2013, 43: 372-381. doi: 10.1016/j.conbuildmat.2013.02.011 [7] KOZUBAL J V, HASSANAT A, TARAWNEH A S, et al. Automatic strength assessment of the virtually modelled concrete interfaces based on shadow-light images[J]. Construction and Building Materials, 2022, 359: 129296. doi: 10.1016/j.conbuildmat.2022.129296 [8] MOHAMAD M E, IBRAHIM I S, ABDULLAH R, et al. Friction and cohesion coefficients of composite concrete-to-concrete bond[J]. Cement and Concrete Composites, 2015, 56: 1-14. [9] COURARD L, PIOTROWSKI T, GARBACZ A. Near-to-surface properties affecting bond strength in concrete repair[J]. Cement and Concrete Composites, 2014, 46: 73-80. doi: 10.1016/j.cemconcomp.2013.11.005 [10] SONG W M, SHU X, HUANG B S, et al. Influence of interface characteristics on the shear performance between open-graded friction course and underlying layer[J]. Journal of Materials in Civil Engineering, 2017, 29(8): 04017077. doi: 10.1061/(ASCE)MT.1943-5533.0001907 [11] ZHANG B S, YU J J, CHEN W Z, et al. Experimental study on bond performance of NC-UHPC interfaces with different roughness and substrate strength[J]. Materials, 2023, 16(7): 2708. doi: 10.3390/ma16072708 [12] VALIKHANI A, JAHROMI A J, MANTAWY IM, et al. Experimental evaluation of concrete-to-UHPC bond strength with correlation to surface roughness for repair application[J]. Construction and Building Materials, 2020, 238: 117753. doi: 10.1016/j.conbuildmat.2019.117753 [13] SHEN Y J, YANG H W, XI J M, et al. A novel shearing fracture morphology method to assess the influence of freeze-thaw actions on concrete-granite interface[J]. Cold Regions Science and Technology, 2020, 169: 102900. doi: 10.1016/j.coldregions.2019.102900 [14] TANG Z P, HUANG F L, PENG H. Effect of 3D roughness characteristics on bonding behaviors between concrete substrate and asphalt overlay[J]. Construction and Building Materials, 2021, 270: 121386. doi: 10.1016/j.conbuildmat.2020.121386 [15] ORESHKIN O, PLATONOV A, PANOV D, et al. Comparison of metrics for shape quality evaluation of textures produced by laser structuring by remelting (waveshape)[J]. Micromachines, 2022, 13(4): 618. doi: 10.3390/mi13040618 [16] CHEN D. Evaluating asphalt pavement surface texture using 3D digital imaging[J]. International Journal of Pavement Engineering, 2020, 21(4): 416-427. doi: 10.1080/10298436.2018.1483503 [17] 张玲, 陈昀灏, 王文涛, 等. 数据驱动的渗蚀-动载作用下粗粒土强度预测[J]. 中国公路学报, 2025, 38(10): 228-239.ZHANG Ling, CHEN Yun-hao, WANG Wen-tao, et al. Strength prediction for coarse-grained soil strength subjected to penetrating erosion and cyclic loading via a data-driven approach[J]. China Journal of Highway and Transport, 2025, 38(10): 228-239. [18] 姚俊峰, 何瑞, 史童童, 等. 基于机器学习的交通流预测方法综述[J]. 交通运输工程学报, 2023, 23(3): 44-67. doi: 10.19818/j.cnki.1671-1637.2023.03.003YAO JUN-feng, HE RUI, SHI TONG-tong, et al. Review on machine learning based traffic flow prediction methods[J]. Journal of Traffic and Transportation Engineering, 2023, 23(3): 44-67. doi: 10.19818/j.cnki.1671-1637.2023.03.003 [19] SADOWSKI Ł, HOŁA J, CZARNECKI S, et al. Pull-off adhesion prediction of variable thick overlay to the substrate[J]. Automation in Construction, 2018, 85: 10-23. doi: 10.1016/j.autcon.2017.10.001 [20] 谭忆秋, 肖神清, 熊学堂. 路面抗滑性能检测与预估方法综述[J]. 交通运输工程学报, 2021, 21(4): 32-47. doi: 10.19818/j.cnki.1671-1637.2021.04.002TAN YI-qiu, XIAO SHEN-qing, XIONG XUE-tang, et al. Review on detection and prediction methods for pavement skid resistance[J]. Journal of Traffic and Transportation Engineering, 2021, 21(4): 32-47. doi: 10.19818/j.cnki.1671-1637.2021.04.002 [21] 曲世琦, 梁尊东, 张鑫. 基于增强极限梯度提升算法的沥青混合料动态模量和相位角预测方法[J]. 科学技术与工程, 2025, 25(3): 1225-1234.QU Shi-qi, LIANG Zun-dong, ZHANG Xin. Enhanced XGBoost-based prediction method for dynamic modulus and phase angle of asphalt mixture[J]. Science Technology and Engineering, 2025, 25(3): 1225-1234. [22] 崔建勋, 要甲, 赵泊媛. 基于深度学习的短期交通流预测方法综述[J]. 交通运输工程学报, 2024, 24(2): 50-64. doi: 10.19818/j.cnki.1671-1637.2024.02.003CUI Jian-xun, YAO Jia, ZHAO Po Yuan. Review on short-term traffic flow prediction methods based on deep learning[J]. Journal of Traffic and Transportation Engineering, 2024, 24(2): 50-64. doi: 10.19818/j.cnki.1671-1637.2024.02.003 [23] 王亦兵, 余宏鑫, 靳凤悦, 等. 连续流动态模型引导的深度学习高速公路交通状态估计[J]. 交通运输工程学报, 2025, 12: 1-31.WANG YI-bing, YU HONG-xin, JIN FENG-yue, et al. Freeway traffic state estimation using continuum-flow-model guided deep learning[J]. Journal of Traffic and Transportation Engineering, 2025, 12: 1-31. [24] 马泽超, 刘小明, 夏汗青, 等. 基于图神经网络的路面病害态势预测方法[J]. 浙江大学学报(工学版), 2024, 58(12): 2596-2608.MA Ze-chao, LIU Xiao-ming, XIA Han-qing, et al. Pavement distress situation prediction method based on graph neural network[J]. Journal of Zhejiang University (Engineering Science), 2024, 58(12): 2596-2608. [25] 邹政, 陈江, 郎洪, 等. 基于机器学习方法的路面病害检测研究综述[J]. 交通信息与安全, 2025, 43(2): 154-168.ZOU Zheng, CHEN Jiang, LANG Hong, et al. A review of pavement distress detection based on machine learning methods[J]. Journal of Transport Information and Safety, 2025, 43(2): 154-168. [26] 曹玉贵, 韦嘉豪, 梅支耀, 等. 基于混合机器学习算法的FRP筋与UHPC粘结强度预测[J]. 沈阳建筑大学学报(自然科学版), 2025, 41(3): 289-298.CAO Yu-gui, WEI Jia-hao, MEI Zhi-yao, et al. Prediction of bond strength between FRP bars and UHPC using a hybrid machine learning algorithm[J]. Journal of Shenyang Jianzhu University (Natural Science), 2025, 41(3): 289-298. [27] 刘文博, 杨喜娟, 王力, 等. 基于SSA-RBFNN的钢管混凝土界面粘结强度研究[J]. 中国安全生产科学技术, 2025, 21(3): 148-155.LIU Wen-bo, YANG Xi-juan, WANG Li, et al. Study on interfacial bond strength of concrete-filled steel tube based on SSA-RBFNN[J]. Journal of Safety Science and Technology, 2025, 21(3): 148-155. [28] ZHANG C, SHEN S H, HUANG H, et al. In-situ dynamic modulus prediction for asphalt pavement combining machine learning algorithm and sensing technology[J]. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(8): 8695-8704. doi: 10.1109/TITS.2024.3385649 [29] VAN DAO D, NGUYEN N L, LY H B, et al. Cost-effective approaches based on machine learning to predict dynamic modulus of warm mix asphalt with high reclaimed asphalt pavement[J]. Materials, 2020, 13(15): 3272. doi: 10.3390/ma13153272 [30] FENG D C, WANG W J, MANGALATHU S, et al. Interpretable XGBoost-SHAP machine-learning model for shear strength prediction of squat RC walls[J]. Journal of Structural Engineering, 2021, 147(11): 04021173. doi: 10.1061/(ASCE)ST.1943-541X.0003115 [31] MEDDAGE D P P, EKANAYAKE I U, WEERASURIYA A U, et al. Explainable Machine Learning (XML) to predict external wind pressure of a low-rise building in urban-like settings[J]. Journal of Wind Engineering and Industrial Aerodynamics, 2022, 226: 105027. doi: 10.1016/j.jweia.2022.105027 [32] RANJBARAN G, RECUPERO D R, ROY C K, et al. C-SHAP: A hybrid method for fast and efficient interpretability[J]. Applied Sciences, 2025, 15(2): 672. doi: 10.3390/app15020672 [33] WANG H J, LIANG Q X, HANCOCK J T, et al. Feature selection strategies: A comparative analysis of SHAP-value and importance-based methods[J]. Journal of Big Data, 2024, 11(1): 44. doi: 10.1186/s40537-024-00905-w [34] XING C, HUI B, WANG H N, et al. Research on the roughness characteristics of cement concrete base after treatment based on 3D laser detection technology[J]. Construction and Building Materials, 2025, 460: 139818. doi: 10.1016/j.conbuildmat.2024.139818 [35] LIANG J, GU X Y, CHEN Y Z, et al. A novel pavement mean texture depth evaluation strategy based on three-dimensional pavement data filtered by a new filtering approach[J]. Measurement, 2020, 166: 108265. doi: 10.1016/j.measurement.2020.108265 [36] MAHATA S, KAR R, MANDAL D. Optimal modelling of (1+α) order butterworth filter under the CFE framework[J]. Fractal and Fractional, 2020, 4(4): 55. doi: 10.3390/fractalfract4040055 [37] 徐星星, 黄昶. 基于高完备性自建数据集的集装箱锁销识别[J]. 华东师范大学学报(自然科学版), 2025(4): 28-37.XU Xing-xing, HUANG Chang. Container lock pins recognition based on a self-built dataset with high completeness[J]. Journal of East China Normal University (Natural Science), 2025(4): 28-37. [38] DUNLOP P, SMITH S. Estimating key characteristics of the concrete delivery and placement process using linear regression analysis[J]. Civil Engineering and Environmental Systems, 2003, 20(4): 273-290. doi: 10.1080/1028660031000091599 [39] YU H, WANG Q F, SHI J Y. Data augmentation generated by generative adversarial network for small sample datasets clustering[J]. Neural Processing Letters, 2023, 55(6): 8365-8384. doi: 10.1007/s11063-023-11315-z [40] LIN L Y, ZHAO S X, ZHANG Y R, et al. Purposive data augmentation strategy and lightweight classification model for small sample industrial defect dataset[J]. IEEE Transactions on Industrial Informatics, 2024, 20(9): 11475-11484. doi: 10.1109/TII.2024.3404053 [41] FENG K, MA S X, XI H Y, et al. Small-sample-data augmentation and transfer strategies for forest cover change monitoring[J]. Ecological Indicators, 2025, 178: 113870. doi: 10.1016/j.ecolind.2025.113870 [42] GAO S, YANG W H, XU M L, et al. U-MLP: MLP-based ultralight refinement network for medical image segmentation[J]. Computers in Biology and Medicine, 2023, 165: 107460. doi: 10.1016/j.compbiomed.2023.107460 [43] NIAZKAR M, MENAPACE A, BRENTAN B, et al. Applications of XGBoost in water resources engineering: A systematic literature review[J]. Environmental Modelling & Software, 2024, 174: 105971. [44] QIU Y G, ZHOU J, KHANDELWAL M, et al. Performance evaluation of hybrid WOA-XGBoost, GWO-XGBoost and BO-XGBoost models to predict blast-induced ground vibration[J]. Engineering with Computers, 2022, 38(5): 4145-4162. [45] HUANG J D, KUMAR G S, REN J L, et al. Accurately predicting dynamic modulus of asphalt mixtures in low-temperature regions using hybrid artificial intelligence model[J]. Construction and Building Materials, 2021, 297: 123655. doi: 10.1016/j.conbuildmat.2021.123655 [46] ZHANG Y, XU X J. Modulus of elasticity predictions through LSBoost for concrete of normal and high strength[J]. Materials Chemistry and Physics, 2022, 283: 126007. doi: 10.1016/j.matchemphys.2022.126007 [47] 李芊昀, 王自溪, 李杰, 等. 基于广义相加模型的北京市2018—2022年地面臭氧气象影响要素研究[J]. 环境科学学报, 2024, 44(4): 206-214.LI Qian-yun, WANG Zi-xi, LI Jie, et al. Research on the influence of meteorological factors on ground-level ozone in Beijing from 2018 to 2022 based on generalized additive model[J]. Acta Scientiae Circumstantiae, 2024, 44(4): 206-214. [48] 吴国超, 李爱莲, 解韶峰. 基于改进的黑翅鸢优化算法-轻梯度提升机建立的转炉炼钢终点温度预测模型[J]. 材料与冶金学报, 2026, 25(1): 37-45.WU Guo-chao, LI Ai-lian, XIE Shao-feng. Improved black-winged kite optimization algorithm-light gradient boosting machine model to predict converter steelmaking endpoint temperature[J]. Journal of Materials and Metallurgy, 2026, 25(1): 37-45. [49] ADLER J, RIBAK E N. Simulated annealing in application to telescope phasing[J]. Physica A: Statistical Mechanics and Its Applications, 2021, 572: 125900. doi: 10.1016/j.physa.2021.125900 [50] ZHANG K, MIN Z H, HAO X T, et al. Enhancing understanding of asphalt mixture dynamic modulus prediction through interpretable machine learning method[J]. Advanced Engineering Informatics, 2025, 65: 103111. doi: 10.1016/j.aei.2025.103111 [51] ZHANG F, CANNONE FALCHETTO A, WANG D, et al. Prediction of asphalt rheological properties for paving and maintenance assistance using explainable machine learning[J]. Fuel, 2025, 396: 135319. doi: 10.1016/j.fuel.2025.135319 [52] PHUNG B N, LE T H, MAI H T, et al. Advancing basalt fiber asphalt concrete design: A novel approach using gradient boosting and metaheuristic algorithms[J]. Case Studies in Construction Materials, 2023, 19: 02528. [53] KENNEDY D M, VAHEY J, HANNEY D. Micro shot blasting of machine tools for improving surface finish and reducing cutting forces in manufacturing[J]. Materials and Design, 2005, 26(3): 203-208. doi: 10.1016/j.matdes.2004.02.013 [54] LU X H, JIA Z Y, ZHANG H X, et al. Tool point frequency response prediction for micromilling by receptance coupling substructure analysis[J]. Journal of Manufacturing Science and Engineering, 2017, 139(7): 071004. doi: 10.1115/1.4035491 [55] LI L, WANG K C P, LI Q, et al. Impacts of sample size on calculation of pavement texture indicators with 1mm 3D surface data[J]. Periodica Polytechnica Transportation Engineering, 2017, 46(1): 42. doi: 10.3311/PPtr.9587 [56] QIAN J S, CEN Y B, et al. Spectrum parameters for runway roughness based on statistical and vibration analysis[J]. International Journal of Pavement Engineering, 2022, 23(11): 3757-3769. doi: 10.1080/10298436.2021.1916821 [57] WONG T T, YEH P Y. Reliable accuracy estimates from k-fold cross validation[J]. IEEE Transactions on Knowledge and Data Engineering, 2020, 32(8): 1586-1594. doi: 10.1109/TKDE.2019.2912815 -
计量
- 文章访问数: 21
- HTML全文浏览量: 5
- PDF下载量: 2
- 被引次数: 0
