引用本文:戴玉峰,苏圣超,崔文霞,汪义旺.基于改进交互式多模型算法的车辆高精度定位[J].控制理论与应用,2025,42(3):590~600.[点击复制]
DAI Yu-feng,SU Sheng-chao,CUI Wen-xia,WANG Yi-wang.High-precision vehicle positioning based on improved interacting multiple model algorithm[J].Control Theory and Technology,2025,42(3):590~600.[点击复制]
基于改进交互式多模型算法的车辆高精度定位
High-precision vehicle positioning based on improved interacting multiple model algorithm
摘要点击 28  全文点击 2  投稿时间：2023-08-28  修订日期：2024-11-03
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2024.30583
  2025,42(3):590-600
中文关键词  车辆定位  交互式多模型  卡尔曼滤波  状态估计
英文关键词  vehicle positioning  interacting multiple model  Kalman filters  state estimation
基金项目  国家自然科学基金项目(61603241), 江苏高校“青蓝工程”(苏教师函[2023]27号), 苏州市科技计划项目(SZS2022015), 苏州市智慧能源技术重点 实验室开放课题基金(SLKSET2308)资助.
作者单位E-mail
戴玉峰 上海工程技术大学 daiyvfeng@163.com 
苏圣超* 上海工程技术大学 jnssc@sues.edu.cn 
崔文霞 上海工程技术大学  
汪义旺 苏州市智慧能源技术重点实验室  
中文摘要
      针对传统交互式多模型算法在车辆运动过程中模型匹配不及时、定位精度较低的问题,本文提出一种结合改进交互式多模型与容积卡尔曼滤波的算法,以改善车辆定位效果.首先,将惯性测量单元和路侧单元的观测结果融合为量测信息;然后,设计一种自适应转弯模型,应对角速度非固定时单一匀速转弯模型无法有效定位车辆的情况;进一步考虑模型非线性、状态向量维度较高的特点,采用容积卡尔曼滤波估计车辆状态;最后,提出改进的交互式多模型算法,通过二次交互优化模型概率.仿真实验表明,本文所提算法可以有效提高模型切换速度和车辆定位的准确性与稳定性,其定位误差相比传统交互式多模型算法降低了8.6%.
英文摘要
      In response to the problem of untimely model matching and low positioning accuracy of the traditional interacting multiple model algorithm during vehicle maneuvering, this paper proposes an algorithm that combines the improved interacting multiple model with cubature Kalman filter to improve the performance of vehicle positioning. Firstly, the observations from inertial measurement unit and road side units are fused into the measurement information. Secondly, an adaptive turn model is designed to cope with the situation that a single constant turn model cannot effectively locate the vehicle when the angular velocity is not fixed. Then, the vehicle state is estimated by cubature Kalman filter, due to the nonlinearity of the model and the high dimensionality of the state vector. Finally, an improved interacting multi model algorithm is proposed to optimize the model probability through twice interactions. The simulation experiments show that the algorithm proposed in this paper can effectively improve the model switching speed, as well as the accuracy and stability of vehicle positioning, and its positioning error is reduced by 8.6% compared with the traditional interacting multiple model algorithm.