基于改进双重无迹卡尔曼滤波算法的车辆状态估计

费明哲¹，王健^1*，于金鹏²，杨君¹，杜若飞¹，王云靖¹，邓欢¹

1.山东交通学院汽车工程学院，山东 济南  250357；2.青岛大学自动化学院，山东 青岛  266071

摘要：针对车辆行驶过程中的状态和参数估计问题，基于软件MATLAB中车辆三自由度动力学模型，分别采用无迹卡尔曼滤波（unscented Kalman filter，UKF）算法、双重无迹卡尔曼滤波（dual unscented Kalman filter，DUKF）算法及采用奇异值分解的改进双重无迹卡尔曼滤波（singular value decomposition-dual unscented Kalman filter，SVD-DUKF）算法，估计车辆在同一工况下的状态及参数。结果表明：UKF算法能保证一定的估计精度，但需时刻输入准确的车身质量等参数，在车辆行驶过程中难以实现；DUKF算法与SVD-DUKF算法有相近的估计精度，但DUKF算法采用Cholesky分解，在车辆运行过程中难以保证误差协方差矩阵为正定矩阵；SVD-DUKF算法更适合估计车辆行驶状态及参数，估计精度较高，适用性较强。

关键词：状态估计；UKF算法；DUKF算法；奇异值分解

Vehicle state estimation based on improved dual unscented Kalman filter

FEI Mingzhe¹, WANG Jian^1*, YU Jinpeng², YANG Jun¹, DU Ruofei¹, WANG Yunjing¹, DENG Huan¹

1.School of Automotive Engineering, Shandong Jiaotong University, Jinan 250357, China

2.School of Automation, Qingdao University, Qingdao 266071, China

Abstract:Based on the 3-DOF dynamics model of the vehicle in MATLAB software, the unscented Kalman filter(UKF), the dual unscented Kalman filter(DUKF) and the improved dual unscented Kalman filter(SVD-DUKF) with singular value decomposition are used to estimate the state and parameters of the vehicle under the same operating conditions respectively. The results show that the UKF algorithm can achieve high estimation accuracy, but the premise is to input accurate parameters such as body mass at all times, which is difficult to reach during the operation of vehicle. The estimation accuracy of DUKF algorithm is similar to SVD-DUKF algorithm, but the former is difficult to guarantee that the error covariance matrix is a positive definite matrix during the operation of vehicle because it uses Cholesky decomposition. The latter with higher estimation accuracy and applicability is more suitable for estimating the driving state and parameters of a vehicle.

Keywords:state estimation; UKF algorithm; DUKF algorithm; singular value decomposition