基于机器学习的高速铁路级配碎石压实质量主控特征与预测研究

PDF(4367 KB)

振动与冲击 ›› 2024, Vol. 43 ›› Issue (21) : 319-328.

论文

基于机器学习的高速铁路级配碎石压实质量主控特征与预测研究

肖宪普1，李新志1，谢康2，郝哲睿2，邓志兴2，李泰灃3

作者信息 +

Main control features and prediction of compaction quality of graded crushed stones for high-speed railway subgrade based on machine learning

XIAO Xianpu1, LI Xinzhi1, XIE Kang2, HAO Zherui2, DENG Zhixing2, LI Taifeng3

Author information +

文章历史 +

摘要

完善高速铁路路基级配碎石压实标准对实现振动压实质量高精度智能预测具有重要意义。首先，开展振动压实试验，基于多参数协同测试方法，探究级配碎石最大干密度ρdmax确定方法；其次，在大量试验数据基础上建立级配碎石特征与ρdmax之间的关系，并采用灰色关联度分析算法明晰影响ρdmax的主控特征；最后，将级配碎石主控特征作为输入特征建立预测ρdmax的机器学习(Machine Learning, ML)模型，并基于ML模型预测性能三层次评价方法确定最优ML模型。结果表明：力学参数动刚度Krb曲线“拐点”对应的压实时间Tlp为级配碎石最佳振动时间，进一步通过Tlp确定级配碎石ρdmax；明晰影响级配碎石ρdmax的主控特征为最大粒径dmax，级配参数b、m，扁平细长颗粒Qe以及洛杉矶磨耗LAA；综合三层次优选结果，各ML模型综合评价指标CEI由小到大分别为：Artificial Neural Network(ANN)模型(1.8797)、Support Vector Regression(SVR)模型(2.9646)、Random Forest(RF)模型(4.5040)、Ridge Regression(Ridge)模型(6.2394)和Decision Tree(DT)模型(7.1319)，ANN模型预测性能最优。研究成果可为高速铁路路基压实质量控制提供新标准，并对路基智能施工提供理论指导。

Abstract

To achieve high precise and intelligent prediction of vibratory compaction quality, it is crucial to refine the compaction standards of graded gravel in high-speed-railway subgrades. Firstly, based on multi-parameter testing, vibration compaction experiments were conducted to explore the determination method of the maximum dry density ρdmax of graded gravel. Secondly, the relationship between characteristics of graded gravel and ρdmax was established based on the extensive experimental data, and the Grey Relational Analysis algorithm was employed to reveal the key controlling characteristics influencing ρdmax. Finally, the key controlling characteristics of graded gravel were used as input features to establish a machine learning (ML) model for predicting ρdmax. And the optimal ML model was determined based on a three-level evaluation method for the predictive performance of the ML model. The results indicated that the compaction time Tlp corresponding to the “inflection point” of the mechanical parameter dynamic stiffness Krb curve represents the optimal vibration time for graded gravel. It was revealed that the key controlling characteristics influencing ρdmax of graded gravel are the maximum particle size dmax, gradation parameters b and m, elongated flat particles Qe, and Los Angeles abrasion LAA. Considering the comprehensive results of the three-level ML model optimization, the calculated comprehensive evaluation index (CEI) for each model are as follows: Artificial Neural Network (ANN) model (1.8797), Support Vector Regression (SVR) model (2.9646), Random Forest (RF) model (4.5040), Ridge Regression (Ridge) model (6.2394), and Decision Tree (DT) model (7.1319). Hence, the predictive performance of the ANN model was optimal. The research results can provide new standards for high-speed railway subgrade compaction quality control and offer theoretical guidance for the intelligent construction of subgrades.

导出引用

肖宪普1, 李新志1, 谢康2, 郝哲睿2, 邓志兴2, 李泰灃3. 基于机器学习的高速铁路级配碎石压实质量主控特征与预测研究[J]. 振动与冲击, 2024, 43(21): 319-328

XIAO Xianpu1, LI Xinzhi1, XIE Kang2, HAO Zherui2, DENG Zhixing2, LI Taifeng3. Main control features and prediction of compaction quality of graded crushed stones for high-speed railway subgrade based on machine learning[J]. Journal of Vibration and Shock, 2024, 43(21): 319-328

参考文献

[1] 马源,方周,韩涛,等. 路基智能压实关键控制参数动态仿真及演变规律[J]. 中南大学学报(自然科学版), 2021, 52(7): 2246-2257.
MA Yuan, FANG Zhou, HAN Tao ,et al. Dynamic simulation and evolution of key control parameters for intelligent compaction of subgrade[J]. Journal of Central South University (Science and Technology), 2021, 52(7): 2246-2257.
[2] 叶阳升,朱宏伟,尧俊凯,等. 高速铁路路基振动压实理论与智能压实技术综述[J]. 中国铁道科学, 2021, 42(5): 1-11.
YE Yangsheng, ZHU Hongwei, YAO Junkai, et al. Review on Vibration Compaction Theory and Intelligent Compaction Technology of High-Speed Railway Subgrade[J]. China Railway Science, 2021, 42(5): 1-11.
[3] 谢康,苏谦,陈晓斌,等. 无砟轨道聚氨酯碎石防水联结层单元模型试验研究[J]. 岩土力学, 2023, 44(8): 1-11.
XIE Kang , SU Qian , CHEN Xiao-bin, et al. Dynamic unit model test research on polyurethane crushed stone waterpro of bonding layer of ballastless track[J]. Rock and Soil Mechanics, 2023, 44(8): 1-11.
[4] 刘先峰,潘申鑫,袁胜洋,等. 压实红层泥岩填料强度与刚度软化和衰减特性研究[J]. 铁道科学与工程学报, 2022, 19(9): 2629-2636.
LIU Xianfeng, PAN Shenxin, YUAN Shengyang, et al. Study on stiffness softening and attenuation characteristics of compacted red mudstone[J]. Journal of Railway Science and Engineering, 2022, 19(9): 2629-2636.
[5] 叶阳升,陈晓斌,惠潇涵,等. 高速铁路路基B组填料振动压实参数优化室内试验研究[J]. 铁道科学与工程学报, 2021, 18(10): 2497-2505.
YE Yangsheng, CHEN Xiaobin, HUI Xiaohan, et al. Laboratory investigation on parameter optimization of vibrating compaction for high-speed railway's Group B[J]. Journal of Railway Science and Engineering, 2021, 18(10): 2497-2505.
[6] 朱俊高,轩向阳,薄以霆. 表面振动压实仪法测定粗粒土密度的影响因素[J]. 水利水运工程学报, 2013(2): 15-19.
ZHU Jungao, XUAN Xiangyang, BO Yiting. Influence factors of dry density of coarse-grained soil measured by surface vibrating compactor[J]. Hydro-Science and Engineering, 2013(2): 15-19.
[7] 王萌,于群丁,肖源杰,等. 透水性基床级配碎石填料宏细观压实特性试验研究[J]. 岩石力学与工程学报, 2022, 41(8): 1701-1716.
WANG Meng, YU Qunding, XIAO Yuanjie, et al. Experimental investigation of macro-and meso-scale compaction characteristics of unbound permeable base materials[J]. Chinese Journal of Rock Mechanics and Engineering, 2022, 41(8): 1701-1716.
[8] 张家玲,徐光辉,蔡英. 连续压实路基质量检验与控制研究[J]. 岩土力学, 2015, 36(4): 1141-1146.
ZHANG Jialing, XU Guanghui, CAI Ying. An investigation on quality inspection and control for continuously compacting subgrade[J]. Rock and Soil Mechanics, 2015, 36(4): 1141-1146.
[9] HU W, POLACZYK P, JIA X, et al. Visualization and quantification of lab vibratory compacting process for aggregate base materials using accelerometer[J]. Transportation Geotechnics, 2020, 25: 100393.
[10] QUAN T V, QUOC D V, SI H L. Evaluating compressive strength of concrete made with recycled concrete aggregates using machine learning approach[J]. Construction and Building Materials, 2022, 323: 126578.
[11] ISIK F, OZDEN G. Estimating compaction parameters of fine- and coarse-grained soils by means of artificial neural networks[J]. Environmental Earth Sciences, 2013, 69(7): 2287-2297.
[12] JAYAN J, SANKAR N. Prediction of Compaction Parameters of Soils using Artificial Neural Network[J], 2015.
[13] WANG X, DONG X, ZHANG Z, et al. Compaction quality evaluation of subgrade based on soil characteristics assessment using machine learning[J]. Transportation Geotechnics, 2022, 32: 100703.
[14] XIE K, CHEN X B, LI T F, et al. A framework for determining the optimal moisture content of high-speed railway-graded aggregate materials based on the lab vibration compaction method[J]. Construction and Building Materials, 2023, 392: 131764.
[15] 谢康, 陈晓斌, 尧俊凯,等. 高铁路基填料振动压实试验参数标准化方法与应用研究[J]. 岩石力学与工程学报, 2023, 42(7): 1799-1810.
XIE Kang, CHEN Xiaobin, YAO Junkai, et al.Study on standardization method and application of vibration compaction test parameters of high-speed railway subgrade filler[J]. Chinese Journal of Rock Mechanics and Engineering, 2023, 42(7): 1799-1810.
[16] FAN Y, LIU C, WANG J. Prediction algorithm for springback of frame-rib parts in rubber forming process by incorporating Sobol within improved grey relation analysis[J]. Journal of Materials Research and Technology, 2021, 13: 1955-1966.
[17] WU E L, ZHU J G, CHEN G, et al. Experimental study of effect of gradation on compaction properties of rockfill materials[J]. Bulletin of Engineering Geology and the Environment, 2020, 79(6): 2863-2869.
[18] ARMAGHANI D J, ASTERIS P G. A comparative study of ANN and ANFIS models for the prediction of cement-based mortar materials compressive strength[J]. Neural Computing and Applications, 2021, 33(9): 4501-4532.
[19] JALAL M, GRASLEY Z, GURGANUS C, et al. A new nonlinear formulation-based prediction approach using artificial neural network (ANN) model for rubberized cement composite[J]. Engineering with Computers, 2022, 38(1): 283-300.
[20] ESCRIBANO Á, WANG D. Mixed random forest, cointegration, and forecasting gasoline prices[J]. International Journal of Forecasting, 2021, 37(4): 1442-1462.
[21] HAN H, SHI B, ZHANG L. Prediction of landslide sharp increase displacement by SVM with considering hysteresis of groundwater change[J]. Engineering Geology, 2021, 280: 105876.
[22] LATIF S D. Developing a boosted decision tree regression prediction model as a sustainable tool for compressive strength of environmentally friendly concrete[J]. Environmental Science and Pollution Research, 2021, 28(46): 65935-65944.
[23] REN X, YANG B, LUO N, et al. The Prediction of Sinter Drums Strength Using Hybrid Machine Learning Algorithms[J]. Computational Intelligence and Neuroscience, 2022, 2022: 4790736.
[24] 宋美, 葛玉辉, 刘举胜. 基于协同进化的动态双重自适应改进PSO算法[J]. 计算机工程与应用, 2020, 56(13): 54-62.
SONG Mei, GE Yuhui, LIU Jusheng.Improved Dynamic Dual Adaptive PSO Algorithm Based on Theory of Co-evolution[J]. Computer Engineering and Applications, 2020, 56(13): 54-62.
[25] ZENG Z Y, ZHU Z Y, YAO W, et al. Accurate prediction of concrete compressive strength based on explainable features using deep learning[J]. Construction and Building Materials, 2022, 329.
[26] CHENG Y, ZHOU W H, XU T. Tunneling-induced settlement prediction using the hybrid feature selection method for feature optimization[J]. Transportation Geotechnics, 2022, 36: 100808.
[27] ZHANG P, WU H N, CHEN R P, et al. Hybrid meta-heuristic and machine learning algorithms for tunneling-induced settlement prediction: A comparative study[J]. Tunnelling and Underground Space Technology, 2020, 99: 103383.
[28] HAMIDIAN P, ALIDOUST P, GOLAFSHANI E M, et al. Introduction of a novel evolutionary neural network for evaluating the compressive strength of concretes: A case of Rice Husk Ash concrete[J]. Journal of Building Engineering, 2022, 61.
[29] ROSS R D, SHI X, CARAM M E V, et al. Veridical causal inference using propensity score methods for comparative effectiveness research with medical claims[J]. Health Services and Outcomes Research Methodology, 2021, 21(2): 206-228.
[30] ZHANG Y, ZHAO Y, SHEN X, et al. A comprehensive wind speed prediction system based on Monte Carlo and artificial intelligence algorithms[J]. Applied Energy, 2022, 305: 117815.