| 407 | 0 | 325 |
| 下载次数 | 被引频次 | 阅读次数 |
为了提升沥青路面车辙预测的精度，通过整合图神经网络与因果推断方法，对多变量时间序列的长短时间模式以及变量之间的相互依赖进行建模，提出了端到端的多元时间序列预测模型。该模型由全局特征提取模块、局部特征提取模块、因果推断模块和双通图卷积模块4个模块组成。在全局特征提取模块，利用注意力机制及门控循环单元（gated recurrent unit,GRU）模型去捕获变量内部长时间模式特征；在局部特征提取模块，使用扩张卷积的方式，用不同大小的卷积神经网络（convolutional neural networks,CNN）卷积核去捕获变量内部不同维度的短时间模式特征；在因果推断模块，采用基于信息理论的因果分析来获取相关性，在使用传递熵来表述变量间因果关系的基础上，整合关系系数矩阵，通过权重配比的方式呈现变量之间的相关性揭示变量间的复杂关系；在双通图卷积模块，将高通滤波器引入传统低通图卷积神经网络，构建双通图卷积预测模块，从低通和高通双通道去同时捕获节点信号或特征的低频分量与高频分量，以提升模型的预测精度。最后，采用交通运输部公路科学研究院轨道数据集RIOHTrack，将提出的模型与经典统计学模型VARIMA、浅层学习模型SVR、深度学习模型GRU、结合注意力机制的GRU和TE-GCN等基准模型进行比较。结果表明，该模型在所有沥青路面结构类别上都取得了最佳的预测性能，且与传统统计学模型相比，基于深度学习的模型更加有效稳定，添加注意力机制的GRU模块可以捕获长期依赖性，从而使模型获得更好的预测性能。该模型为沥青路面的车辙预测提供了一种高效的方法，有望用于未来路面结构和养护方案设计，以提升路面使用寿命。
Abstract：To enhance the prediction accuracy of asphalt pavement rutting, this study introduces an end-to-end multivariate time series prediction model that integrates graph neural networks(GNN) with causal inference methodologies.The proposed model aims to effectively capture long-term and short-term temporal patterns as well as interdependencies among multiple variables. The model comprises four modules: global feature extraction, local feature extraction,causal inference, and dual-channel graph convolution. The global feature extraction module employs attention mechanisms and gated recurrent units(GRU) to capture long-term temporal dependencies within variables. The local feature extraction module utilizes dilated convolutional neural networks(CNN) with various kernel sizes to extract short-term temporal patterns at different scales. In the causal inference module, relationships among variables are identified using transfer entropy based on information theory, resulting in a relationship coefficient matrix that quantifies complex dependencies among variables. The dual-channel graph convolution module extends traditional low-pass graph convolutional neural networks by integrating a high-pass filter, simultaneously capturing low-frequency and high-frequency components of node signals or features to potentially improve prediction accuracy. The proposed approach was evaluated using the RIOHTrack dataset from the Research Institute of Highway Track, with comparisons conducted against several benchmark models, including the classical statistical model VARIMA, shallow learning model SVR, deep learning model GRU, attention mechanism-enhanced GRU, and TE-GCN. Experimental results indicate that the proposed model achieves superior predictive performance across various categories of asphalt pavement structures. Compared to traditional statistical models, deep learning-based models are more effective and stable, and the GRU module enhanced with attention mechanisms can capture long-term dependencies, further enhancing predictive performance. Overall, the proposed model provides a potentially effective solution for predicting asphalt pavement rutting and may offer practical insights for future pavement structure design and maintenance planning aimed at extending pavement lifespan.
[1] HARDOON D R, SZEDMAK S, SHAWE-TAYLOR J.Canonical correlation analysis:an overview with application to learning methods[J]. Neural Computation, 2004, 16(12):2639-2664.
[2] HAN M, REN W J. Global mutual information-based feature selection approach using single-objective and multiobjective optimization[J]. Neurocomputing, 2015, 168:47-54.
[3] RESHEF D N, RES HEF Y A, FINUCANE H K, et al.Detecting novel associations in large data sets[J]. Science,2011, 334(6062):1518-1524.
[4] SHI J, DING Z H, LEE W J, et al. Hybrid forecasting model for very-short term wind power forecasting based on grey relational analysis and wind speed distribution features[J]. IEEE Transactions on Smart Grid, 2014, 5(1):521-526.
[5] LIEBSCHER E. Copula-based dependence measures[J].Dependence Modeling, 2014, 2(1):49-64.
[6] SUN Y Q, LI J Y, LIU J X, et al. Using causal discovery for feature selection in multivariate numerical time series[J]. Machine Learning, 2015, 101(1):377-395.
[7]任伟杰，韩敏.多元时间序列因果关系分析研究综述[J].自动化学报，2021, 47(1):64-78.REN W J, HAN M. Survey on causality analysis of multivariate time series[J]. Acta Automatica Sinica, 2021, 47(1):64-78.(in Chinese)
[8] GRANGER C W J. Investigating causal relations by econometric models and cross-spectral methods[J]. Econometrica, 1969, 37(3):424-438.
[9] BARNETT L, SETH A K. The MVGC multivariate granger causality toolbox:a new approach to granger-causal inference[J]. Journal of Neuroscience Methods, 2014, 223:50-68.
[10] BROWN S F, BRUNTON J M. An introduction to the analytical design of bituminous pavements[R]. England:University of Nottingham, 1992.
[11] MILLER J S, BELLINGER W Y. Distress identification manual for the long-term pavement performance program[R]. United States:Federal Highway Administration, 2003.
[12] WANG X C, ZHAO J, LI Q Q, et al. A hybrid model for prediction in asphalt pavement performance based on support vector machine and grey relation analysis[J]. Journal of Advanced Transportation, 2020(1):7534970.
[13]张金喜，郭旺达，宋波，等.基于随机森林的沥青路面性能预测[J].北京工业大学学报，2021, 47(11):1256-1263.ZHANG J X, GUO W D, SONG B, et al. Performance prediction of asphalt pavement based on random forest[J].Journal of Beijing University of Technology, 2021, 47(11):1256-1263.(in Chinese)
[14] MARCELINO P, DE LUEDES ANTUNES M, FORTUNATO E, et al. Machine learning approach for pavement performance prediction[J]. International Journal of Pavement Engineering, 2021, 22(3):341-354.
[15]赵港，王千阁，姚烽，等.大规模图神经网络系统综述[J].软件学报，2022, 33(1):150-170.ZHAO G, WANG Q G, YAO F, et al. Survey on largescale graph neural network systems[J]. Journal of Software,2022, 33(1):150-170.(in Chinese)
[16] RIBEIRO L F R, SAVERESE P H P, FIGUEIREDO D R.Struc2vec:learning node representations from structural identity[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, August 13-17, 2017, Halifax, NS, Canada. New York:Association for Computing Machinery, 2017:385-394.
[17] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[EB/OL].(2016-09-09)[2024-02-20]. https：//arxiv.org/abs/1609.02907v4.
[18] CHEN J, MA T F, XIAO C. FastGCN:fast learning with graph convolutional networks via importance sampling[EB/OL].(2018-01-30)[2024-02-20]. https：//arxiv.org/abs/1801.10247v1.
[19] HUANG X, LI J D, HU X. Accelerated attributed network embedding[M]//Proceedings of the 2017 SIAM International Conference on Data Mining. Philadelphia, PA:Society for Industrial and Applied Mathematics, 2017:633-641.
[20] HAMILTON W L, YING R, LESKOVEC J. Inductive representation learning on large graphs[EB/OL].(2017-06-07)[2024-02-20]. https：//arxiv.org/abs/1706.02216v4.
[21]胡堰翔.基于特征融合的多元时间序列预测方法研究[D].成都：电子科技大学，2021.HU Y X. Research on multivariate time series forecasting method based on feature fusion[D]. Chengdu:University of Electronic Science and Technology of China, 2021.(in Chinese)
[22] SCHREIBER T. Measuring information transfer[J]. Physical Review Letters, 2000, 85(2):461-464.
[23] DU S D, LI T R, YANG Y, et al. Multivariate time series forecasting via attention-based encoder-decoder framework[J]. Neurocomputing, 2020, 388:269-279.
[24] ANSARIZADEH F, TAY D B, THIRUVADY D, et al.Augmenting graph convolutional neural networks with highpass filters[C]//Proceedings of Joint IAPR International Workshops:Structural, Syntactic, and Statistical Pattern Recognition, S+SSPR 2020, January 21-22, 2021, Padua,Italy. Cham:Springer International Publishing, 2021:77-86.
[25] VYAS V, SINGH A P, SRIVASTAVA A. Prediction of asphalt pavement condition using FWD deflection basin parameters and artificial neural networks[J]. Road Materials and Pavement Design, 2021, 22(12):2748-2766.
[26] ZHANG J R, CAO J D, HUANG W, et al. Rutting prediction and analysis of influence factors based on multivariate transfer entropy and graph neural networks[J]. Neural Networks, 2023, 157:26-38.
基本信息:
DOI：10.12194/j.ntu.20240427001
中图分类号:U418.68
引用信息:
[1]陈凯,王小荷,时欣利等.基于因果推断的图神经网络沥青路面车辙预测方法[J].南通大学学报(自然科学版),2025,24(01):18-27+50.DOI:10.12194/j.ntu.20240427001.
基金信息:
国家重点研发计划项目（2020YFA0714300）; 综合交通运输理论交通运输行业重点实验室（南京现代综合交通实验室）开放课题资助项目（MTF2023004）