汽车工程 ›› 2026, Vol. 48 ›› Issue (3): 663-675.doi: 10.19562/j.chinasae.qcgc.2026.03.016
• • 上一篇
曹守启1,高雅琪1,周国峰1(
),陈渐伟2,周志松3,姜加胜1
收稿日期:2025-03-14
修回日期:2025-06-09
出版日期:2026-03-25
发布日期:2026-03-19
通讯作者:
周国峰
E-mail:gfzhou@shou.edu.cn
基金资助:
Shouqi Cao1,Yaqi Gao1,Guofeng Zhou1(
),Jianwei Chen2,Zhisong Zhou3,Jiasheng Jiang1
Received:2025-03-14
Revised:2025-06-09
Online:2026-03-25
Published:2026-03-19
Contact:
Guofeng Zhou
E-mail:gfzhou@shou.edu.cn
摘要:
高精度车辆侧倾动力学建模对提升车辆主动安全控制性能至关重要。然而,车辆系统具有强非线性与参数不确定性,基于传统机理分析的建模方法难以准确估计侧倾角和侧倾角速度。针对此问题,本文提出一种融合多尺度卷积神经网络与Transformer的数据建模方法。该模型利用多尺度卷积核提取含噪声数据中的多频域特征,提高模型对噪声干扰的鲁棒性;同时,结合Transformer的注意力机制,有效捕捉侧倾动力学中的长时序依赖关系,进一步提升建模精度。为验证模型性能,本研究基于CarSim高保真仿真平台和实车道路实验数据,将其与传统Transformer、LSTM、GRU等数据驱动模型以及物理模型进行对比分析。实验结果表明,所提出的CNN-Transformer混合模型在侧倾角和侧倾角速度预测任务中表现最优,预测决定系数R2 均高于0.974 5,实现了对车辆侧倾动力学的准确建模。
曹守启,高雅琪,周国峰,陈渐伟,周志松,姜加胜. 基于CNN-Transformer的车辆侧倾动力学建模及实验验证[J]. 汽车工程, 2026, 48(3): 663-675.
Shouqi Cao,Yaqi Gao,Guofeng Zhou,Jianwei Chen,Zhisong Zhou,Jiasheng Jiang. Modeling and Experimental Validation of Vehicle Roll Dynamics Based on CNN-Transformer[J]. Automotive Engineering, 2026, 48(3): 663-675.
表1
不同尺度卷积核和时间步的侧倾角与侧倾角速度预测结果误差对比"
| 卷积核类别 | 时间步类别 | 侧倾角 | 侧倾角速度 | |||||
| R2 | MAE | RMSE | R2 | MAE | RMSE | |||
| 卷积核为3 | 时间步为1 | 0.913 0 | 0.004 4 | 0.005 2 | 0.845 1 | 0.004 6 | 0.005 9 | |
| 卷积核为5 | 0.983 4 | 0.001 6 | 0.002 3 | 0.980 7 | 0.001 8 | 0.002 1 | ||
| 卷积核为7 | 0.960 9 | 0.002 4 | 0.003 5 | 0.980 4 | 0.001 7 | 0.002 1 | ||
| 多尺度卷积 | 0.997 5 | 0.000 7 | 0.000 9 | 0.993 4 | 0.001 0 | 0.001 2 | ||
| 多尺度卷积 | 时间步为2 | 0.958 5 | 0.002 6 | 0.003 6 | 0.941 0 | 0.003 0 | 0.003 7 | |
表2
仿真工况侧倾角预测结果误差对比"
| 模型 | 8字仿真工况 | 随机仿真工况 | 高速蛇行仿真工况 | ||||||||
| R2 | MAE | RMSE | R2 | MAE | RMSE | R2 | MAE | RMSE | |||
| Physical Model | 0.845 4 | 0.008 1 | 0.009 2 | 0.846 3 | 0.007 0 | 0.008 5 | 0.366 2 | 0.017 8 | 0.019 6 | ||
| LSTM | 0.983 7 | 0.002 7 | 0.003 0 | 0.984 3 | 0.002 4 | 0.002 7 | 0.878 9 | 0.008 0 | 0.008 6 | ||
| GRU | 0.991 9 | 0.001 5 | 0.002 1 | 0.983 7 | 0.002 5 | 0.002 8 | 0.918 6 | 0.006 6 | 0.007 0 | ||
| Transformer | 0.991 9 | 0.001 7 | 0.002 1 | 0.985 5 | 0.001 7 | 0.002 6 | 0.997 4 | 0.000 9 | 0.001 3 | ||
| CNN-Transformer | 0.999 3 | 0.000 5 | 0.000 6 | 0.992 9 | 0.001 4 | 0.001 8 | 0.998 6 | 0.000 7 | 0.000 9 | ||
表3
仿真工况侧倾角速度预测结果误差对比"
| 模型 | 8字仿真工况 | 随机仿真工况 | 高速蛇行仿真工况 | ||||||||
| R2 | MAE | RMSE | R2 | MAE | RMSE | R2 | MAE | RMSE | |||
| Physical Model | 0.682 9 | 0.004 4 | 0.006 6 | 0.635 2 | 0.007 2 | 0.011 4 | 0.363 7 | 0.024 8 | 0.028 5 | ||
| LSTM | -8.067 9 | 0.027 8 | 0.034 5 | -6.287 | 0.045 9 | 0.051 0 | 0.466 1 | 0.022 2 | 0.026 1 | ||
| GRU | -34.032 5 | 0.067 5 | 0.069 6 | -14.74 | 0.072 5 | 0.075 0 | 0.293 9 | 0.025 6 | 0.030 0 | ||
| Transformer | 0.954 7 | 0.001 8 | 0.002 5 | 0.966 1 | 0.002 3 | 0.003 5 | 0.993 9 | 0.001 8 | 0.002 8 | ||
| CNN-Transformer | 0.984 5 | 0.001 0 | 0.001 5 | 0.993 3 | 0.001 1 | 0.001 5 | 0.997 2 | 0.001 3 | 0.001 9 | ||
表4
实车工况侧倾角预测结果误差对比"
| 模型 | 8字实车工况 | 随机实车工况 | |||||
| R2 | MAE | RMSE | R2 | MAE | RMSE | ||
| Physical Model | 0.686 1 | 0.006 5 | 0.007 3 | 0.686 1 | 0.006 5 | 0.007 3 | |
| LSTM | 0.927 6 | 0.002 8 | 0.003 5 | 0.770 4 | 0.006 9 | 0.008 5 | |
| GRU | 0.904 0 | 0.003 0 | 0.004 0 | 0.721 8 | 0.007 8 | 0.009 3 | |
| Transformer | 0.981 0 | 0.001 4 | 0.001 8 | 0.990 0 | 0.001 4 | 0.001 8 | |
| CNN-Transformer | 0.992 5 | 0.000 9 | 0.001 1 | 0.997 5 | 0.000 7 | 0.000 9 | |
表5
实车工况侧倾角速度预测结果误差对比"
| 模型 | 8字实车工况 | 随机实车工况 | |||||
| R2 | MAE | RMSE | R2 | MAE | RMSE | ||
| Physical Model | 0.609 3 | 0.003 4 | 0.004 2 | 0.609 3 | 0.003 4 | 0.004 2 | |
| LSTM | 0.442 8 | 0.004 1 | 0.005 0 | 0.692 9 | 0.007 0 | 0.008 3 | |
| GRU | 0.140 9 | 0.005 1 | 0.006 3 | 0.602 6 | 0.007 9 | 0.009 5 | |
| Transformer | 0.963 0 | 0.000 9 | 0.001 3 | 0.992 3 | 0.001 0 | 0.001 3 | |
| CNN-Transformer | 0.983 3 | 0.000 6 | 0.000 9 | 0.993 4 | 0.001 0 | 0.001 2 | |
| [1] | 金智林, 何少炜, 黄舒伟. 融合侧翻稳定性的SUV换道轨迹规划方法研究[J]. 机械工程学报, 2023, 59(4): 145-154. |
| JIN Z L, HE S W, HUANG S W. Research on SUV lane changing trajectory planning method integrating rollover stability[J]. Journal of Mechanical Engineering, 2023, 59(4): 145-154. | |
| [2] | GÁSPÁR P, NÉMETH B. Integrated control design for driver assistance systems based on LPV methods[J]. International Journal of Control, 2016, 89(12): 2420-2433. |
| [3] | RYU Y I, KANG D O, HEO S J, et al. Rollover mitigation for a heavy commercial vehicle[J]. International Journal of Automotive Technology, 2010, 11(2): 283-287. |
| [4] | DAHMANI H, PAGÈS O, EL HAJJAJI A, et al. Observer-based robust control of vehicle dynamics for rollover mitigation in critical situations[J]. IEEE Transactions on Intelligent Transportation Systems, 2014, 15(1): 274-284. |
| [5] | 金智林, 翁建生, 胡海岩. 汽车侧翻及稳定性分析[J]. 机械科学与技术, 2007, 26(3): 355-358. |
| JIN Z L, WENG J S, HU H Y. Analysis of vehicle′s rollover stability[J]. Mechanical Science and Technology for Aerospace Engineering, 2007, 26(3): 355-358. | |
| [6] | 金智林, 翁建生, 胡海岩. 汽车侧翻预警及防侧翻控制[J]. 动力学与控制学报, 2007, 5(4): 365-369. |
| JIN Z L, WENG J S, HU H Y. Rollover warning and anti-rollover control for automobiles[J]. Journal of Dynamics and Control, 2007, 5(4): 365-369. | |
| [7] | ATAEI M, KHAJEPOUR A, JEON S. Model predictive rollover prevention for steer-by-wire vehicles with a new rollover index[J]. International Journal of Control, 2020, 93(1): 140-155. |
| [8] | ATAEI M, KHAJEPOUR A, JEON S. Development of a novel general reconfigurable vehicle dynamics model[J]. Mechanism and Machine Theory, 2021, 156: 104147. |
| [9] | PAN Y J, NIE X B, LI Z X, et al. Data-driven vehicle modeling of longitudinal dynamics based on a multibody model and deep neural networks[J]. Measurement, 2021, 180: 109541. |
| [10] | ZHOU Z S, WANG Y F, ZHOU G F, et al. Vehicle lateral dynamics-inspired hybrid model using neural network for parameter identification and error characterization[J]. IEEE Transactions on Vehicular Technology, 2024, 73(11): 16173-16186. |
| [11] | GARIMELLA G, FUNKE J, WANG C, et al. Neural network modeling for steering control of an autonomous vehicle[C]. 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2017: 2609-2615. |
| [12] | 胡明辉, 朱广曜, 刘长贺, 等. 考虑迟滞特性的卡尔曼滤波和门控循环单元神经网络的锂离子电池SOC联合估计[J]. 汽车工程, 2023, 45(9): 1688-1701. |
| HU M H, ZHU G Y, LIU C H, et al. Joint estimation of state of charge for lithium-ion battery with Kalman filtering and gated recurrent unit neural networks considering hysteresis characteristics[J]. Automotive Engineering, 2023, 45(9): 1688-1701. | |
| [13] | 陈渐伟, 于传强, 刘志浩, 等. 多轴特种车辆的数据建模方法及横向动力学应用[J]. 兵工学报, 2023, 44(1): 165-175. |
| CHEN J W, YU C Q, LIU Z H, et al. Data modeling of multi-axle special vehicles and lateral dynamics applications[J]. Acta Armamentarii, 2023, 44(1): 165-175. | |
| [14] | CHEN J W, YU C Q, WANG Y F, et al. Hybrid modeling for vehicle lateral dynamics via AGRU with a dual-attention mechanism under limited data[J]. Control Engineering Practice, 2024, 151: 106015. |
| [15] | 杨志刚, 李俣静, 夏超, 等. 基于LSTM-POD的汽车湍流尾迹的高时间分辨速度场重构[J]. 汽车工程, 2024, 46(7): 1302-1313. |
| YANG Z G, LI Y J, XIA C, et al. A high time-resolution reconstruction on the automotive turbulent wake based on LSTM-POD[J]. Automotive Engineering, 2024, 46(7): 1302-1313. | |
| [16] | 陈吉清, 冯雨佳, 兰凤崇, 等. 基于1dCNN-LSTM量化单体异常性的动力电池故障诊断方法[J]. 汽车工程, 2024, 46(7): 1177-1188. |
| CHEN J Q, FENG Y J, LAN F C, et al. Fault diagnosis method for power battery based on quantification of cell abnormality with 1dCNN-LSTM[J]. Automotive Engineering, 2024, 46(7): 1177-1188. | |
| [17] | 方培俊, 蔡英凤, 陈龙, 等. 基于ED-LSTM的智能汽车神经网络横向动力学建模与控制[J]. 力学学报, 2022, 54(7): 1896-1908. |
| FANG P J, CAI Y F, CHEN L, et al. Neural network lateral dynamics modeling and control based on ed-lstm for intelligent vehicle[J]. Chinese Journal of Theoretical and Applied Mechanics, 2022, 54(7): 1896-1908. | |
| [18] | MENG D L, LI Z X, CHU H Q, et al. Transformer-based sideslip angle estimation for autonomous vehicle applications[C]. 2024 14th Asian Control Conference. IEEE, 2024: 226-231. |
| [19] | ANIK B M T H, ISLAM Z, ABDEL-ATY M. A time-embedded attention-based transformer for crash likelihood prediction at intersections using connected vehicle data[J]. Transportation Research Part C: Emerging Technologies, 2024, 169: 104831. |
| [20] | 胡杰, 陈琳, 王志红, 等. 基于Transformer的纯电动汽车充电时间预测[J]. 汽车工程, 2024, 46(11): 2059-2067. |
| HU J, CHEN L, WANG Z H, et al. Transformer-based prediction of charging time for pure electric vehicles[J]. Automotive Engineering, 2024, 46(11): 2059-2067. | |
| [21] | WANG J T, ZHANG X L, DONG Y W, et al. Roll stability control of in-wheel motors drive electric vehicle on potholed roads[J]. Control Engineering Practice, 2025, 157: 106247. |
| [22] | MENG Q H, ZHENG H Y, CHEN C C, et al. Active shimmy control method for driverless electric vehicle considering unknown sensor measurement error and nonlinearities[J]. International Journal of Control, Automation and Systems, 2023, 21(7): 2246-2258. |
| [23] | CHEN C J, LIU X Y, YANG L, et al. A novel tracking estimation algorithm for the detection beam’s roll angle[J]. Journal of the Brazilian Society of Mechanical Sciences and Engineering, 2024, 46(9): 575. |
| [24] | PARSONS T, SEO J, KIM B, et al. Dust de-filtering in LiDAR applications with conventional and CNN filtering methods[J]. IEEE Access, 2024, 12: 22032-22042. |
| [25] | 朱继洪, 裴继红, 赵阳. 卷积神经网络(CNN)训练中卷积核初始化方法研究[J]. 信号处理, 2019, 35(4): 641-648. |
| ZHU J H, PEI J H, ZHAO Y. Research on convolution kernel initialization method in convolutional neural network (CNN) training[J]. Journal of Signal Processing, 2019, 35(4): 641-648. | |
| [26] | 冯天培, 孙跃东, 王岩松, 等. 车内综合烦躁度评价的时序平滑激励级谱CNN模型[J]. 汽车工程, 2020, 42(6): 784-792. |
| FENG T P, SUN Y D, WANG Y S, et al. Annoyance evaluation model of vehicle interior noise based on time-series smoothed excitation level spectrum CNN model[J]. Automotive Engineering, 2020, 42(6): 784-792. | |
| [27] | 余志生. 汽车理论[M]. 2版. 北京: 机械工业出版社, 1990.YU Z S. Automobile theory[M]. 2nd ed. Beijing: China Machine Press, 1990. |
| [28] | GENG M S, LI J Y, XIA Y J, et al. A physics-informed Transformer model for vehicle trajectory prediction on highways[J]. Transportation Research Part C: Emerging Technologies, 2023, 154: 104272. |
| [29] | AIZPURUA J I, PEÑA-ALZOLA R, OLANO J, et al. Probabilistic machine learning aided transformer lifetime prediction framework for wind energy systems[J]. International Journal of Electrical Power & Energy Systems, 2023, 153: 109352. |
| [1] | 张曦,杨颖,陈超君,王春风,杨磊. 增强双流Transformer的柴油发动机剩余寿命预测模型[J]. 汽车工程, 2025, 47(2): 292-300. |
| [2] | 祝琴,韩沈阳,曾明如,赖平红,吴垂茂,胡玮轶. SFW-YOLOv8复杂场景视频车辆检测模型[J]. 汽车工程, 2024, 46(12): 2290-2302. |
| [3] | 赵健,李文旭,朱冰,张培兴,汤瑞,李嘉胜. 面向智能驾驶测试的可变跟驰特性交通车建模方法[J]. 汽车工程, 2024, 46(11): 1952-1961. |
| [4] | 胡杰,陈琳,王志红,卿海华,王浩杰. 基于Transformer的纯电动汽车充电时间预测[J]. 汽车工程, 2024, 46(11): 2059-2067. |
| [5] | 赵霞,李朝,付锐,葛振振,王畅. 基于深度卷积-Tokens降维优化视觉Transformer的分心驾驶行为实时检测[J]. 汽车工程, 2023, 45(6): 974-988. |
|
||