吉林大学学报(工学版) ›› 2025, Vol. 55 ›› Issue (9): 2858-2863.doi: 10.13229/j.cnki.jdxbgxb.20241285

• 车辆工程·机械工程 • 上一篇    

基于机器学习的汽车设计智能拟合方法

兰巍1(),周政1,王冠宇2,王伟3(),张苗苗1   

  1. 1.吉林大学 汽车底盘集成与仿生全国重点实验室,长春 130022
    2.一汽-大众汽车有限公司 技术开发造型中心,长春 130011
    3.一汽-大众汽车有限公司 技术开发部,成都 610100
  • 收稿日期:2024-11-30 出版日期:2025-09-01 发布日期:2025-11-14
  • 通讯作者: 王伟 E-mail:lanwei@jlu.edu.cn;wangwei_gps@126.com
  • 作者简介:兰巍(1980-),女,教授,博士.研究方向:运载工具创新与智能化设计.E-mail:lanwei@jlu.edu.cn
  • 基金资助:
    吉林省重大科技项目(GF-2022-03883);国家重点研发计划项目(ZkjcD105181350452131154)

Intelligent fitting method for vehicle design based on machine learning

Wei LAN1(),Zheng ZHOU1,Guan-yu WANG2,Wei WANG3(),Miao-miao ZHANG1   

  1. 1.National Key Laboratory of Automotive Chassis Integration and Bionics,Jilin University,Changchun 130022,China
    2.Technical Development and Styling Center,FAW-Volkswagen Automobile Co. ,Ltd. ,Changchun 130011,China
    3.Technical Development Department,FAW-Volkswagen Automobile Co. ,Ltd. ,Chengdu 610100,China
  • Received:2024-11-30 Online:2025-09-01 Published:2025-11-14
  • Contact: Wei WANG E-mail:lanwei@jlu.edu.cn;wangwei_gps@126.com

摘要:

为提升车企工作效率、为设计师提供设计框架,有效提升设计的准确性和客观性,提出了一种利用机器学习拟合汽车设计前期工作、后期个性覆盖件设计的方法。该方法主要基于扩散模型和文本反转技术,首先对用户画像进行分类并训练模型;其次,借助已有市场反馈度较高的成型车对应的用户画像进行高度拟合。由此可通过调节用户画像数据等信息,便可生成前期定义车型效果图、汽车覆盖套件等。与现有流程相比,该方法能够高度拟合设计前期工作,为设计师提供更直观、更客观的前期定义车型。因此,该方法在汽车设计、市场调研、造型策略制定、创意草案生成、覆盖件设计等领域具有广泛的应用前景。

关键词: 车辆工程, 汽车设计, 用户画像, 扩散模型, 人工智能, 文本反转

Abstract:

In order to improve the work efficiency of automobile enterprises, provide the design framework for designers, and effectively improve the accuracy and objectivity of the design, a method of using machine learning was proposed to fit the preliminary work of vehicle design and the later personalized cover parts design. This method is mainly based on diffusion models and text inversion technology. Firstly, the user profile was classified and the model was trained. Secondly, a high-degree fitting with the help of the existing corresponding user portrait of the molding car with high market feedback was performed. By adjusting user profile data and other information, pre defined vehicle renderings, car coverage kits, etc. can be generated. Compared with existing processes, the proposed method can highly fit the pre design work and provide designers with more intuitive and objective pre definition of vehicle models. Therefore, this method has broad application prospects in fields such as automotive design, market research, styling strategy formulation, creative draft generation, and panel design.

Key words: vehicle engineering, vehicle design, user portrait, diffusion models, artificial intelligence, text inversion technology

中图分类号: 

  • TP18

图1

扩散模型的原理图"

图2

文本反转的模型原理图"

图3

基于扩散模型和文本反转的模型框架图"

表1

3种汽车造型出图方法的比较"

人工扩散模型扩散模型+文本反转
出图效率
美学稳定性
[1] 彭剑, 项娇. 整车产品开发流程浅析[J]. 上海汽车,2010(8): 43-47, 51.
Peng Jian, Xiang Jiao. Brief analysis of the vehicle product development process[J]. Shanghai Automobile, 2010(8): 43-47, 51.
[2] 张洪志. 汽车整车设计开发流程优化研究[J]. 汽车测试报告, 2023(13): 13-15.
Zhang Hong-zhi. Research on the optimization of the complete vehicle design and development process[J]. Automotive Test Report, 2023(13): 13-15.
[3] 王庆, 赵发珍. 基于“用户画像”的图书馆资源推荐模式设计与分析[J]. 现代情报, 2018, 38(3): 105-109, 137.
Wang Qing, Zhao Fa-zhen. Design and analysis of library resource recommendation model based on "User portrait"[J]. Modern Intelligence, 2018, 38(3): 105-109, 137.
[4] 刘海, 卢慧, 阮金花, 等. 基于“用户画像”挖掘的精准营销细分模型研究[J]. 丝绸, 2015, 52(12): 37-42, 47.
Liu Hai, Lu Hui, Ruan Jin-hua, et al. Research on precision marketing segmentation model based on "user portrait" mining[J]. Silk, 2015, 52(12): 37-42, 47.
[5] Mostafa M M, El-Masry A A. Citizens as consumers: profiling e-government services' users in Egypt via data mining techniques[J].International Journal of Information Management, 2013, 33(4): 627-641.
[6] de Andres J, Pariente B, Gonzalez-Rodriguez M, et al.Towards an automatic user profiling system for online information sites Identifying demographic determining factors[J].Online Information Review, 2015,39(1): 61-80.
[7] 李映坤. 大数据背景下用户画像的统计方法实践研究[D]. 北京: 首都经济贸易大学统计学院, 2016.
Li Ying-kun. Research on the practical statistical methods of user portrait in the context of big data[D]. Beijing: School of Statistics, Capital University of Economics and Business, 2016.
[8] 张小可, 沈文明, 杜翠凤. 贝叶斯网络在用户画像构建中的研究[J]. 移动通信, 2016, 40(22): 22-26.
Zhang Xiao-ke, Shen Wen-ming, Du Cui-feng. Bayesian networks in user portrait construction[J]. Mobile Communications, 2016,40 (22): 22-26.
[9] 李恒超, 林鸿飞, 杨亮, 等. 一种用于构建用户画像的二级融合算法框架[J]. 计算机科学, 2018, 45(1): 157-161.
Li Heng-chao, Lin Hong-fei, Yang Liang, et al. A secondary fusion algorithm framework for constructing user portraits[J]. Computer Science, 2018,45 (1): 157-161.
[10] 林燕霞, 谢湘生. 基于社会认同理论的微博群体用户画像[J]. 情报理论与实践, 2018, 41(3): 142-148.
Lin Yan-xia, Xie Xiang-sheng. User portrait of microblog group based on social identity theory[J]. Information Theory and Practice, 2018,41 (3): 142-148.
[11] Travis D. E-commerce Usability: Tools and Techniques to Perfect the On-line Experience[M]. Boca haton, FL: CRC Press, 2002.
[12] Lafouge T, Lardy J P, Abdallah N B. Improving information retrieval by combining user profile and document segmentation[J]. Information Processing Management An International Journal, 1996(3): 305-315.
[13] Guimarães T P. Perfil de usuários de biblioteca governamental: o caso do Ministério da Saúde[J]. Perspectivas em Ciência da Information, 2007,12(3): 96-115.
[14] 马安华. 基于用户行为分析的精确营销系统设计与实现[D]. 南京: 南京邮电大学计算机学院, 2013.
Ma An-hua. Design and implementation of accurate marketing system based on user behavior analysis[D]. Nanjing: School of Computer Science, Nanjing University of Posts and Telecommunications, 2013.
[15] 李映坤. 大数据背景下用户画像的统计方法实践研究[D]. 北京: 首都经济贸易大学统计学院, 2016.
Li Ying-kun. Practical research on the statistical methods of user portrait in the background of big data[D]. Beijing: School of Statistics, Capital University of Economics and Business, 2016.
[16] 曾建勋. 精准服务需要用户画像[J]. 数字图书馆论坛,2017(12): 1.
Zeng Jian-xun. Accurate service needs user portrait [J]. Digital Library Forum, 2017(12): 1.
[17] 刘海鸥, 孙晶晶, 苏妍嫄, 等. 基于用户画像的旅游情境化推荐服务研究[J]. 情报理论与实践, 2018(10): 87-92.
Liu Hai-ou, Sun Jing-jing, Su Yan-yuan, et al. Research on tourism situational recommendation service based on user portrait[J]. Information Theory and Practice, 2018(10): 87-92.
[18] 刘海鸥, 陈晶, 孙晶晶, 等.面向大数据的移动数字图书馆情境化推荐系统研究[J].图书馆工作与研究,2018(9): 58-64.
Liu Hai-ou, Chen Jing, Sun Jing-jing, et al. Research on contextualized recommendation system for mobile digital libraries in the era of big data [J]. Library Work and Research, 2018(9): 58-64..
[1] 孙天骏,杨惠喆,蔡荣贵,冯嘉仪,冉锐,刘斌. 面向纯电动汽车自适应巡航系统的人性化起停控制策略[J]. 吉林大学学报(工学版), 2025, 55(9): 2847-2857.
[2] 赵俊武,曲婷,胡云峰. 基于自适应采样的智能车辆轨迹规划方法[J]. 吉林大学学报(工学版), 2025, 55(8): 2802-2816.
[3] 于贵申,陈鑫,唐悦,赵春晖,牛艾佳,柴辉,那景新. 激光表面处理对铝-铝粘接接头剪切强度的影响[J]. 吉林大学学报(工学版), 2025, 55(8): 2555-2569.
[4] 高金武,孙少龙,王舜尧,高炳钊. 基于电机转矩补偿的增程器转速波动抑制策略[J]. 吉林大学学报(工学版), 2025, 55(8): 2475-2486.
[5] 孙佩铭,王喆. 基于导向差分进化算法的党务活动调度优化方法[J]. 吉林大学学报(工学版), 2025, 55(8): 2761-2770.
[6] 贾美霞,胡建军,肖凤. 基于多软件联合的车用电机变工况多物理场仿真方法[J]. 吉林大学学报(工学版), 2025, 55(6): 1862-1872.
[7] 宋学伟,于泽平,肖阳,王德平,袁泉,李欣卓,郑迦文. 锂离子电池老化后性能变化研究进展[J]. 吉林大学学报(工学版), 2025, 55(6): 1817-1833.
[8] 肖纯,易子淳,周炳寅,张少睿. 基于改进鸽群优化算法的燃料电池汽车模糊能量管理策略[J]. 吉林大学学报(工学版), 2025, 55(6): 1873-1882.
[9] 李伟东,马草原,史浩,曹衡. 基于分层强化学习的自动驾驶决策控制算法[J]. 吉林大学学报(工学版), 2025, 55(5): 1798-1805.
[10] 卢荡,索艳茹,孙宇航,吴海东. 基于无量纲格式的轮胎侧倾侧偏力学特性预测[J]. 吉林大学学报(工学版), 2025, 55(5): 1516-1524.
[11] 高镇海,郑程元,赵睿. 真实与虚拟场景下自动驾驶车辆的主动安全性验证与确认综述[J]. 吉林大学学报(工学版), 2025, 55(4): 1142-1162.
[12] 张涛,林黄达,余中军. 混合动力车辆换挡的实时滚动优化控制方法[J]. 吉林大学学报(工学版), 2025, 55(4): 1215-1224.
[13] 袁杰,王军博,陈歆,黄馨,张傲翔,崔安琪. 人工智能在超高性能混凝土中的应用研究进展[J]. 吉林大学学报(工学版), 2025, 55(3): 771-789.
[14] 卢荡,王晓凡,吴海东. TWEEL轮胎接地压力均布特性分析[J]. 吉林大学学报(工学版), 2025, 55(3): 811-819.
[15] 陈鑫,张祥源,武子涛,于贵申,杨立飞. 工艺顺序对车用铝薄板胶-PFSSW接头拉剪性能的影响[J]. 吉林大学学报(工学版), 2025, 55(2): 468-475.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!