汽车工程 ›› 2022, Vol. 44 ›› Issue (12): 1910-1918.doi: 10.19562/j.chinasae.qcgc.2022.12.012

所属专题: 底盘&动力学&整车性能专题2022年

• • 上一篇    下一篇

基于二分法的车辆状态参数融合估计

查云飞1(),吕小龙1,刘鑫烨1,马芳武1,2   

  1. 1.福建工程学院,福建省汽车电子与电驱动技术重点实验室,福州  350118
    2.吉林大学,汽车仿真与控制国家重点实验室,长春  130025
  • 收稿日期:2022-06-05 修回日期:2022-07-15 出版日期:2022-12-25 发布日期:2022-12-22
  • 通讯作者: 查云飞 E-mail:zhayf@fjut.edu.cn
  • 基金资助:
    国家自然科学基金(51675057);福建省科技重大专项(2020HZ03018)

Fusion Estimation of Vehicle State Parameters Based on Dichotomy

Yunfei Zha1(),Lü Xiaolong1,Xinye Liu1,Fangwu Ma1,2   

  1. 1.Fujian University of Technology,Fujian Key Laboratory of Automotive Electronics and Electric Drive,Fuzhou  350118
    2.Jilin University,State Key Laboratory of Automotive Simulation and Control,Changchun  130025
  • Received:2022-06-05 Revised:2022-07-15 Online:2022-12-25 Published:2022-12-22
  • Contact: Yunfei Zha E-mail:zhayf@fjut.edu.cn

摘要:

为提高车辆状态参数估计的精度和可靠性,提出一种基于二分法的车辆状态参数融合估计方法。首先,设计了基于车辆3自由度动力学模型的扩展卡尔曼滤波算法和由数据驱动的径向基神经网络车辆状态参数估计算法。然后,为了进一步提高估计算法的可靠性和减小单一算法的估计误差,提出将模型驱动的估计算法和数据驱动的估计算法相补偿的融合估计方法,基于二分法设置扩展卡尔曼滤波和径向基神经网络估计结果的权重,利用估计算法的融合提高估计精度。最后通过MATLAB/Simulink与CarSim的联合仿真和实车在环试验对该融合方法的有效性进行了验证。结果表明,估计结果变化趋势与实际相符,所提出的融合算法的估计精度比单一扩展卡尔曼滤波算法和径向基神经网络算法有明显的提升。

关键词: 状态参数, 融合估计, 二分法, 扩展卡尔曼滤波算法, 径向基神经网络算法

Abstract:

In order to enhance the accuracy and reliability of vehicle state parameters estimation, a vehicle state fusion estimation method based on dichotomy is proposed. Firstly, an extended Kalman filtering (EKF) algorithm based on the vehicle 3-DOF dynamic model and the data-driven radial basis function (RBF) neural network estimation algorithm are designed. Then for further raising the reliability of the estimation algorithm and reducing the estimation error of single algorithm, a fusion estimation method is proposed with model-driven estimation algorithm and data-driven estimation algorithm compensating each other. The weights of the estimation results of EKF and RBF neural network are assigned based on dichotomy and the fusion of algorithms enhances the estimation accuracy. Finally, the co-simulation of MATLAB/Simulink and CarSim, and the real vehicle in-the-loop experiment are performed to verify the effectiveness of the fusion method. The results show that the change trend of the estimation result is consistent with the real one, and the estimation accuracy of fusion algorithm proposed is significantly enhanced, compared to RBFNN and single EKF algorithms.

Key words: state parameters, fusion estimation, dichotomy, EKF algorithm, RBF neural network algorithm