摘要
在汽车的运行过程中,悬架系统的状态不可避免地会发生改变。为了准确评估悬架参数的长期变化,尤其是实现早期故障预警,提出了一种基于车辆实际行驶状态的悬架参数辨识方法,首先在车辆的关键部位安装振动传感器,采集振动加速度信号。然后,通过递推最小二乘算法对悬架的弹簧刚度和减震器阻尼系数进行初步识别。在此基础上,进一步采用Rao-Blackwellized粒子滤波算法对初步辨识结果进行二次优化。最后,结合实测的车辆硬点坐标和通过辨识得到的悬架参数,基于多体动力学原理构建车辆动力学模型,与实际设计参数进行对比,并进行整车动力学仿真以验证辨识参数的准确性。实验结果表明,该方法在识别悬架弹簧刚度和减震器阻尼系数方面具有很高的精度,与真实值的最大偏差仅为2.50%和1.82%。同时,车辆动力学模型的仿真输出与实测载荷谱的均方根误差控制在5%以内。该方法显著提高了悬架系统参数辨识的精确度,是一种高精度的汽车悬架参数在线辨识算法。
The condition of the suspension system will inevitably change witha vehicle’s operation.To accurately assess the long-term changes in suspension parameters,especially the early fault warning,this paper proposes a suspension parameter identification method based on the actual driving state of the vehicle.First,it installs vibration sensors at key parts of the vehicle to capture vibration acceleration signals.Then,it uses a recursive least squares algorithm to preliminarily identify the spring stiffness and shock absorber damping coefficient of the suspension.On this basis,the Rao-Blackwellized particle filter algorithm is employed to finely optimize the two parameters.Finally,combining the measured vehicle hard point coordinates and the identified suspension parameters,a vehicle dynamics model is constructed based on the principles of multi-body dynamics,and compared with the actual design parameters to verify the accuracy of the identified parameters.Results show the method achieves high accuracy in identifying the spring stiffness and shock absorber damping coefficient of the suspension,with the maximum deviation from the true value being only 2.50%and 1.82%respectively.Meanwhile,the root mean square error of the simulation output of the vehicle dynamics model and the actual measured load spectrum is controlled within 5%.The method markedly improves the accuracy of suspension system parameter identification.
作者
王姝
董传昊
张大伟
赵轩
周辰雨
邵帅
WANG Shu;DONG Chuanhao;ZHANG Dawei;ZHAO Xuan;ZHOU Chenyu;SHAO Shuai(School of Automotive,Chang’an University,Xi’an 710064,China)
出处
《重庆理工大学学报(自然科学)》
北大核心
2025年第7期19-27,共9页
Journal of Chongqing University of Technology:Natural Science
基金
国家自然科学基金面上项目(52172362,52372375)
陕西省重点研发计划项目(2024GX-YBXM-260)
陕西省科技成果转化计划项目(2024CG-CGZH-19)
陕西省自然科学基础研究项目(2022JQ543543)。
关键词
递推最小二乘算法
RBPF算法
实车载荷谱
参数辨识
recursive least squares algorithm
RBPF algorithm
real vehicle load spectrum
parameter identification
作者简介
王姝,女,高级工程师,主要从事电动汽车及其控制技术研究,E-mail:shuwang@chd.edu.cn;通信作者:张大伟,男,博士,讲师,主要从事车辆系统动力学和故障诊断研究,E-mail:dwzhang@chd.edu.cn。
