中文核心期刊
CSCD来源期刊
中国科技核心期刊
RCCSE中国核心学术期刊
重庆交通大学学报（自然科学版）

重庆交通大学学报（自然科学版） ›› 2017, Vol. 36 ›› Issue (5): 115-120.DOI: 10.3969/j.issn.1674-0696.2017.05.20

• 车辆与机电工程 • 上一篇    

基于遗传算法的魔术公式轮胎模型参数两级辨识

边伟1，龚佳慧2，文爱民1，陈林山1，刘奕贯1，2   

  1. （1.南京交通职业技术学院 汽车工程学院，江苏 南京 211188； 2.南京农业大学 工学院，江苏 南京 210031）
  • 收稿日期:2016-03-12 修回日期:2016-11-12 出版日期:2017-05-18 发布日期:2017-05-18
  • 作者简介:边伟（1967—），女，河南商丘人，副教授，主要从事汽车机电一体化技术与运用方面的研究。E-mail：b3415417@126.com。 通信作者：龚佳慧（1991—），女，江苏扬州人，硕士，主要从事车辆电子控制技术应用方面的研究。E-mail：15195955668@163.com。
  • 基金资助:
    国家自然科学基金项目（51175269）

Two Levels of Parameter Identification of Magic Formula Tire Model Based on Genetic Algorithm

BIAN Wei1, GONG Jiahui2, WEN Aimin1, CHEN Linshan1, LIU Yiguan1，2   

  1. (1.College of Automobile Engineering, Nanjing Communications Institute of Technology, Nanjing 211188, Jiangsu, P. R. China; 2.College of Engineering, Nanjing Agricultural University, Nanjing 210031,Jiangsu, P. R. China)
  • Received:2016-03-12 Revised:2016-11-12 Online:2017-05-18 Published:2017-05-18

PDF

摘要： 为了提高魔术公式(Magic formula，MF)轮胎模型参数的辨识精度及速度，采用参数的两级辨识的方法，定义公式相关的参数B，C，D，E为一级参数，轮胎模型的特性参数为二级参数。首先基于Matlab遗传算法工具箱对一级参数进行辨识，然后基于一级参数的辨识结果再次利用遗传算法对二级参数进行辨识，并且将辨识出的参数带入到魔术公式计算轮胎的受力，拟合出不同载荷下的魔术公式轮胎模型的纵向力随滑移率变化曲线。对一级参数辨识的相对残差为2.196 8%，且遗传代数为40代左右就收敛；二级参数的辨识相对残差为0.840 3%，且遗传代数为20代左右时就收敛。辨识结果表明：对魔术公式轮胎模型参数分两级辨识的方法，可以保证参数辨识的精度，并有效提高辨识的效率,为实时参数辨识提供了高效可靠的方法。

关键词: 车辆工程, 轮胎模型, 魔术公式, 遗传算法, 参数辨识

Abstract: To improve the identification accuracy and speed of magic formula (MF) tire model, a method that dividing all parameters into two levels was used. The formula-related parameters B, C, D, E were defined as the first level parameters and the characteristic parameters of the tire model as the second level. Firstly, the first level parameters were identified by Matlab genetic algorithm toolbox, and then the second level parameters were identified by genetic algorithm again based on the former results. At the same time, the identification results were brought into the magic formula to calculate the tire force, and then the curve of longitudinal force of MF tire model changing with slip rate under different loads was fitted. The identification result shows that the relative residual error of first level parameter identification is 2.196 8%, and the result is converged within 40 iterations; the relative residual error of the second level is 0.840 3%, and the result is converged within 20 iterations. It is concluded that the method that two levels of parameter identification of MF tire model based on genetic algorithm can ensure the parameter identification accuracy and improve identification efficiency, which provides an efficient and reliable method for real-time parameter identification.

Key words: vehicle engineering, tire model, magic formula, genetic algorithm, parameter identification

中图分类号: