摘要
为了克服线性系统中卡尔曼滤波发散这一问题,提出了一种基于信息矩阵的AKF.首先,通过对AKF滤波方程的等价推导,针对性解决了协方差矩阵的初始值问题,在信息矩阵初值P-10未知(P0一般初始化为0)时,滤波的稳定性和准确性都不受影响.另外,在基于信息矩阵的AKF基础上,为了更好适应实时场景,一种简化AKF算法(SAKF)被提出.仿真结果表明,基于信息矩阵的AKF与传统AKF(基于Sage-Husa的AKF)相比较,在P0未知时,其目标跟踪滤波效果明显优于传统AKF,该滤波算法在运动目标跟踪、航迹描绘等领域都有广泛的应用前景.
In target tracking based on adaptive Kalman filter（AKF）,the initial value of covariance matrix has great influence on the accuracy and stability of target tracking filtering.On the consequence of the lack of prior information,the initial value of covariance is often unknown.Therefore,it will cause serious filter divergence,which affects the filtering performance.In order to overcome the divergence problem of Kalman filter,an AKF based on information matrix is proposed in this paper.First,by using the equivalent derivation of AKF equation,an new AKF algorithm is developed based on the information matrix,which focuses on solving the initial value of covariance matrix.Especially,when the initial value of the information matrix is unknown（generally,initializes 0）,the stability and accuracy of filter is almost not affected.In addition,on the basis of new AKF,a simple AKF（SAKF）is presented to adapt the real-time scene.Simulation results show that the filter effects surpass traditional AKF（based on Sage-Husa）,which leading to a broad application prospects in the field of the moving object tracking and tracking depicting.
出处
《西南大学学报（自然科学版）》
CAS
CSCD
北大核心
2016年第7期172-178,共7页
Journal of Southwest University(Natural Science Edition)
基金
四川省科技厅基金项目(2015FZ0088)
关键词
自适应卡尔曼滤波器
信息矩阵
滤波发散
目标跟踪
adaptive Kalman filter
information matrix
filter divergence
moving object tracking
