针对水下无人平台搭载单矢量水听器目标跟踪鲁棒性差的问题，提出一种基于L2范数正则化粒子滤波的单站纯方位角度水下目标跟踪算法。该方法引入测量量与预测量误差的规则范数，得到粒子的似然函数，从而平衡测量值与预测值，提高目标跟踪鲁棒性。仿真结果表明，相较于粒子滤波（PF）算法，基于L2范数正则化（L2-RPF）优于基于L1范数正则化（L1-RPF），其目标方位跟踪精度更高，且经过多次Monte Carlo实验得到的L2-RPF目标方位跟踪平均误差和均方根误差均较小。利用水下滑翔机平台搭载单矢量水听器（水下声学滑翔机）在中国南海海域进行探测跟踪的试验数据，对算法性能进行了验证，采用L2-RPF处理得到的目标方位跟踪精度相较于其他算法较高，一定程度上修正了测量野值带来的跟踪误差。
Aiming at the poor target tracking robustness of underwater unmanned platform equipped with a single vector hydrophone, a single-station pure azimuth-angle underwater target tracking algorithm based on L2 norm regularized particle filter is proposed. This method introduces the regular norm of the error between the measured quantity and the predicted quantity, and obtains the likelihood function of the particles, thereby balancing the relationship between the measured value and the predicted value, improving the robustness of target tracking. The simulation results show that compared with the particle filter (PF) algorithm, the target bearing tracking accuracy of L2 norm regularization (L2-RPF) is better than that of L1 norm regularization (L1-RPF); The L2-RPF target bearing tracking average error and root mean square error obtained by Monte Carlo experiment are both small. Using the test data in the South China Sea detected and tracked by the underwater glider platform equipped with single-vector hydrophone (underwater acoustic glider), the algorithm performance was verified, and the target azimuth tracking accuracy obtained by L2-RPF processing is higher compared with others, which corrects the tracking error caused by measurement outliers to a certain extent.
2020,42(12): 111-116 收稿日期：2020-09-18
DOI：10.3404/j.issn.1672-7649.2020.12.022
分类号：TP391
作者简介：田德艳(1989-),女,硕士,助理工程师,研究方向为水下无人平台目标探测与跟踪
参考文献：
[1] 王超, 孙芹东, 兰世泉, 等. 水下声学滑翔机目标探测性能南海海试分析[J]. 声学技术, 2018, 37(6): 149-150
[2] 王文龙, 王超, 韩梅, 等. 矢量水听器在水下滑翔机上的应用研究[J]. 兵工学报, 2019, 40(12): 2580-2587
[3] 王超, 笪良龙, 韩梅, 等. 单矢量水听器的高分辨目标方位跟踪算法研究[J]. 应用声学, 2017, 36(1): 59-66
[4] 田德艳, 张少康, 王超. Kalman滤波估计在水下目标跟踪中的应用[C]//中国声学学会水声学分会2019年学术年会.
[5] 李天成, 范红旗, 孙树栋. 粒子滤波理论、方法及其在多目标跟踪中的应用[J]. 自动化学报, 2015, 41(12): 1981-2002
[6] 宋绪栋, 蔚婧, 李晓花, 等. 基于纯方位角测量的水下目标被动跟踪技术[J]. 水下无人系统学报, 2012, 20(5): 353-358
[7] 张林琳, 杨日杰, 熊华. 非线性滤波方法在水下目标跟踪中的应用[J]. 火力与指挥控制, 2010(8): 17-21
[8] OLIVIER L E, CRAIQ I K. Fault-tolerant nonlinear MPC using particle filtering[J]. Ifac Papersonline, 2016, 49(7): 177-182
[9] 许枫, 纪永强, 郭占军, 等. 基于混合粒子滤波的水下小目标跟踪[J]. 应用声学, 2015(4): 19-24
[10] 章飞, 孙睿. 基于粒子滤波的水下目标被动跟踪算法[J]. 2010, 24(1): 84-87.
[11] 任宇飞, 李宇, 黄海宁. 能量值和方位信息结合的粒子滤波算法[J]. 哈尔滨工程大学学报, 2017(7)
[12] 石桂欣, 鄢社锋, 郝程鹏, 等. 不完全测量下长基线系统的水下目标跟踪算法[J]. 声学学报, 2019
[13] 石桂欣, 鄢社锋, 刘宇. 水下目标跟踪的改进非线性滤波快速算法[J]. 应用声学, 2020, 39(1): 89-96
[14] 张铁栋, 万磊, 王博, 等. 基于改进粒子滤波算法的水下目标跟踪[J]. 上海交通大学学报, 2012, 046(6): 943-948
[15] 王宏建, 徐金龙, 李娟, 等. 非平稳非高斯测量噪声条件下改进差分粒子滤波算法研究[J]. 兵工学报, 2014
[16] GYORGY K, KELEMEN, ANDRAS, et al. Unscented Kalman filters and particle filter methods for nonlinear state estimation[J]. Procedia Technology, 2014, 12: 65-74
[17] ANDREAS S, FREDRIK L, THOMAS B. S. Learning nonlinear State-Space models using smooth Particle-filter-Based likelihood approximations[J]. 2018.
[18] YI W, FU L, ngel F. Garcia-Fernandez, et al Particle filtering based track-before-detect method for passive array sonar system[J]. Signal Processing, 2019: 165
[19] HERBST E, SCHORFTHEIDE F. Tempered particle filtering[J]. PIER Woring Paper Archive, 2016
[20] 金盛龙, 李宇, 黄海宁. 水下多目标方位的联合检测与跟踪[J]. 声学学报, 2019
[21] MUSSO C, OUDJANEN, LEGLAND F. Sequential Monte Carlo methods in practice by ar[M]. New York: Springer-Verlag, 2001.
[22] 刘敏, 陈恩庆, 杨守义. 正则化粒子滤波在水下目标跟踪中的应用[J]. 电视技术, 2012, 36(9)
[23] 唐现国, 何祖军. 一种基于正则粒子滤波器的目标跟踪算法[J]. 舰船科学技术, 2008, 30(4): 135-137
