汽车工程 ›› 2022, Vol. 44 ›› Issue (10): 1547-1555.doi: 10.19562/j.chinasae.qcgc.2022.10.009

所属专题: 底盘&动力学&整车性能专题2022年

• • 上一篇    下一篇

基于无迹卡尔曼滤波和门控循环单元的道路坡度估计

秦大同(),王康,冯继豪,刘永刚,程坤,夏玉   

  1. 重庆大学,机械传动国家重点实验室,重庆  400044
    2.重庆大学机械与运载工程学院,重庆  400044
  • 收稿日期:2022-04-19 修回日期:2022-06-02 出版日期:2022-10-25 发布日期:2022-10-21
  • 通讯作者: 秦大同 E-mail:dtqin@cqu.edu.cn
  • 基金资助:
    国家自然科学基金重点支持项目(U1764259)

Road Slope Estimation Based on Unscented Kalman Filtering and Gated Recurrent Unit

Datong Qin(),Kang Wang,Jihao Feng,Yonggang Liu,Kun Cheng,Yu Xia   

  1. Chongqing University,State Key Laboratory of Mechanical Transmission,Chongqing  400044
    2.College of Mechanical and Vehicle Engineering,Chongqing University,Chongqing  400044
  • Received:2022-04-19 Revised:2022-06-02 Online:2022-10-25 Published:2022-10-21
  • Contact: Datong Qin E-mail:dtqin@cqu.edu.cn

摘要:

针对外接激光雷达等传感器普适性差,而传统道路坡度估计方法仅根据车载CAN总线数据在车辆起步、换挡、制动和停车4种特殊工况中的估计误差较大的问题,提出了一种基于无迹卡尔曼滤波(UKF)和门控循环单元(GRU)的道路坡度估计方法。根据车速等数据识别工况,在非特殊工况下,建立车辆动力学模型并采用UKF来估计坡度;在特殊工况下,将规律性不稳定的时序坡度转换为距序坡度,并利用GRU进行短距坡度预测。仿真和实车试验结果表明:在非特殊工况下,该方法通过UKF可准确估计道路坡度;在特殊工况下,该方法通过GRU可有效跟踪距序坡度变化趋势,显著提高了道路坡度估计精度。

关键词: 坡度估计, 无迹卡尔曼滤波, 门控循环单元, 短距坡度

Abstract:

In view of the poor universality of external lidar and other sensors, and the large error of traditional road slope estimation method based on onboard CAN bus data in four special conditions of vehicle, including starting, gear shifting, braking, and stopping, a road slope estimation approach based on unscented Kalman filter (UKF) and gated recurrent unit (GRU) is proposed. The working condition is determined according to vehicle speed and other data. Under non-special conditions, the vehicle dynamic model is established and UKF is used to estimate the slope. Under special working conditions, the time series slope with unstable-regularity is converted to the distance series slope, and GRU is used to estimate the short-range slope. The results of simulation and real vehicle tests show that under non-special conditions, the method proposed can accurately estimate the road slope using UKF, and under special conditions, the method can effectively track the changing trend of distance series slope through GRU, significantly enhancing the estimation accuracy of road slope.

Key words: slope estimation, unscented Kalman filtering, gated recurrent unit, short-range slope