存在非高斯重尾分布噪声的纯方位目标跟踪算法  

Bearings-Only Target Tracking Algorithm with Non-Gaussian

作　　者：刘灿 王辉[1] 林德福[1] 崔晓曦 徐晗晖 LIU Can;WANG Hui;LIN Defu;CUI Xiaoxi;XU Hanhui(School of Astronautics,Beijing Institute of Technology,Beijing 100081,China;Norinco Group Institute of Navigation and Control Technology,Beijing 100089,China;School of Design and Arts,Beijing Institute of Technology,Beijing 100081,China)

机构地区：[1]北京理工大学宇航学院,北京100081 [2]中国兵器工业导航与控制技术研究所,北京100089 [3]北京理工大学设计与艺术学院,北京100081

出　　处：《兵工学报》2023年第5期1469-1481,共13页Acta Armamentarii

基　　金：国家自然科学基金项目(61827901)。

摘　　要：纯方位目标跟踪是目标跟踪研究中的热点问题,针对目标跟踪方程中的非高斯重尾分布噪声问题,提出了一种针对非高斯重尾分布噪声的卡尔曼滤波算法。该方法通过建立基于存在异常值的高斯分布的层次高斯模型来近似未知的非高斯重尾分布系统过程噪声和测量噪声,并使用变分贝叶斯推断来学习混合概率,解决混合概率不确定带来的滤波性能下降的问题,从而提高滤波的鲁棒性。同时针对纯方位目标跟踪模型的非线性,结合修正增益卡尔曼滤波来降低量测方程非线性的影响。数值仿真结果表明,相对于EKF、UKF和变分贝叶斯卡尔曼滤波PEKF-VB、VBEKF,新算法VBMGEKF估计精度分别提高了69.31%、58.08%、127.84%和9.36%,具备更好的鲁棒性与精度。The bearings-only target tracking is a classic problem in the reserach on target tracking.Focusing on the problem of non-Gaussian heavy-tailed distributed noise in the model of target tracking,this paper proposes a new Kalman filter algorithm.Firstly,the hierarchical Gaussian model is established to approximate the unknown process noise and measurement noise of the non-Gaussian heavy-tailed distributed system.Next,the variational Bayesian inference is used to learn Mixture Probability to solve the problem of the filter s performance degradation caused by the uncertainty of Mixture Probability,so as to improve the robustness of the filter.At the same time,for the nonlinearity of the bearings-only target tracking model,Modified Gain Kalman filter is used to reduce the influence of nonlinearity on the observation equation.The numerical simulations have verified that the proposed filter has better estimation accuracy and robustness than EKF,UKF and the variational Bayesian Kalman filters PEKF-Vb and VBEKF.The estimation accuracy of the proposed algorithm VBMGEKF is improved by 69.31%,58.08%,127.84%and 9.36%.

关 键 词：纯方位目标跟踪 变分贝叶斯 层次高斯模型 重尾分布噪声 修正增益卡尔曼滤波 

分 类 号：V556.6[航空宇航科学与技术—人机与环境工程] TN959.74[电子电信—信号与信息处理]