摘要
为解决车辆运动状态估计中卡尔曼滤波器因量测噪声统计特性发生变化而出现性能降低甚至滤波发散的问题,针对双轨车辆模型提出了一种基于变分贝叶斯自适应容积卡尔曼滤波算法的估计方法。该方法从概率角度将车辆运动状态及量测噪声一起作为待估计的变量,通过定点迭代的方式不断逼近量测噪声的真实后验分布,最终获得局部最优解。基于CarSim与MATLAB/Simulink的联合仿真结果表明:在时变噪声协方差条件下,该算法与CKF相比具有精度高、性能稳定的优点。
In order to solve the performance degradation and even filter divergence problem of Kalman filter caused by the varying measurement noise statistical characteristics in vehicle motion state estimation,an estimation method based on variational Bayesian adaptive cubature Kalman filter is proposed for the dual-track vehicle model.This method takes the vehicle motion state and the measurement noise as the variables and to be estimated together,and continuously approximates the true posterior distribution of the measurement noise through fixed-point iteration,finally obtains the local optimal solution.The simulation results based on the CarSim-MATLAB/Simulink joint platform show that under the condition of Time-variant noise covariance,the algorithm has the advantages of high accuracy and stable performance compared with cubature Kalman filter.
作者
孙坚添
叶贻财
郭聪聪
Sun Jiantian;Ye Yicai;Guo Congcong(School of Automotive and Traffic Engineering,Jiangsu University,Zhenjiang City,Jiangsu Province 212013,China)
出处
《农业装备与车辆工程》
2021年第5期37-41,共5页
Agricultural Equipment & Vehicle Engineering
关键词
容积卡尔曼滤波
变分贝叶斯
车辆运动状态
噪声
cubature Kalman filter
variational Bayesianian
vehicle motion state
noise
