欢迎访问《汽车安全与节能学报》，
汽车安全与节能学报

汽车安全与节能学报 ›› 2025, Vol. 16 ›› Issue (4): 648-656.DOI: 10.3969/j.issn.1674-8484.2025.04.015

• 智能驾驶与智慧交通 • 上一篇    

基于SQP和GRNN的商用客车动力学参数自适应辨识

房熙博(), 宁一高(), 赵轩, 周猛   

  1. 长安大学 汽车学院， 西安 710018， 中国
  • 收稿日期:2025-04-11 修回日期:2025-05-20 出版日期:2025-08-30 发布日期:2025-08-27
  • 通讯作者: *宁一高，讲师。E-mail：ningyigao@chd.edu.cn。
  • 作者简介:房熙博（2000—），男（汉），甘肃，硕士研究生。E-mail：fangxibo@chd.edu.cn。
  • 基金资助:
    国家自然科学基金项目(52402492);国家自然科学基金项目(52372375);中国博士后科学基金资助项目(2023M730358);陕西省自然科学基础研究计划项目(2024JC-YBQN-0564);陕西省科技成果转化项目(2024CG-CGZH-19)

Adaptive identification of dynamic parameters for commercial buses based on SQP and GRNN

FANG Xibo(), NING Yigao(), ZHAO Xuan, ZHOU Meng   

  1. School of Automobile, Chang’an University, Xi’an 710018, China
  • Received:2025-04-11 Revised:2025-05-20 Online:2025-08-30 Published:2025-08-27

RichHTML

7

PDF

46

摘要：

提出了一种基于广义回归神经网络（GRNN）模型和序列二次规划（SQP）算法的自适应辨识策略，用于获取商用客车动力学参数并对其实时辨识。建立GRNN模型，用SQP算法获取GRNN模型的训练集对其进行训练，使其根据车辆的运行状态，自适应辨识出关键参数；搭建TruckSim与Matlab/Simulink联合仿真平台，在不同工况下进行仿真试验。结果表明：相较于固定参数模型，在正弦波转角工况下，采用该模型的质心侧偏角与TruckSim模型的最大值误差减小73.9%；其侧倾角与TruckSim模型的最大值误差减少了76.7%；在双移线工况下，这2个误差分别减小98.0%和63.1%。从而，证明了本文方法的可行性和有效性。

关键词: 汽车安全, 商用客车, 序列二次规划（SQP）算法, 广义回归神经网络（GRNN）模型, 动力学参数, 自适应辨识

Abstract:

An adaptive identification strategy was proposed based on the generalized regression neural network (GRNN) model and the sequential quadratic programming (SQP) algorithm to obtain and identify the key dynamic parameters of commercial vehicles in real time. A GRNN model was established and trained using the training data obtained via the SQP algorithm, with being enabled to adaptively identify key parameters according to the vehicle’s operating states. A co-simulation platform was built with integrating the TruckSim and the Matlab/Simulink to conduct simulation experiments under various driving conditions. The results show that compared with a fixed-parameters model, under the sine wave steering input condition, the maximum error of the vehicle’s sideslip angle is reduced by 73.9% than the TruckSim model with the maximum error of the roll angle being reduced by 76.7%. Meanwhile, these two errors are reduced by 98.0% and 63.1% under the double-lane change condition, respectively. Therefore, these results demonstrate the feasibility and effectiveness of the proposed method.

Key words: vehicle safety, commercial buses, SQP (sequential quadratic programming) algorithm, GRNN (general regression neural network) model, dynamics parameters, adaptive identification

中图分类号: