利用改进单分类支持向量机提升舰船尾流目标的检测准确率

doi:10.3969/j.issn.1000-1093.2020.09.021

摘要/Abstract

摘要： 舰船在航行过程中会在尾部产生一段包含大量气泡的湍流区域，通过对尾流的声学检测可以有效地跟踪船舶。基于一种改进单分类支持向量机（OCSVM）算法，利用无尾流情况下回波信号作为训练集的一个最优分类器，用于尾流回波信号模式判断。对回波信号进行降噪处理，进而提出一种自适应特征提取方法对回波信号进行处理；将特征提取作为输入，使用两层决策边界的双阈值OCSVM算法进行尾流检测。仿真结果表明，与常规OCSVM算法相比，改进算法在不同信噪比下的检测准确率均有提升，检测准确率最高可达96.27%，具有较好的工程应用价值。

关键词: 舰船, 尾流检测, 特征提取, 自适应阈值, 单分类支持向量机

Abstract: A turbulent area containing a large number of bubbles is generated at the tail of the ship during the voyage. The ship can be effectively tracked by the acoustic detection of wake. An improved one-class support vector machine (SVM) algorithm is proposed，which uses the echo signal in the wake-free case as an optimal classifier for the training set for judging the wake echo signal mode. The echo signal is denoised，and then an adaptive feature extraction method is proposed to process the echo signal; a dual-threshold one-class SVM with two-layer decision boundary is used for wake detection by using feature extraction as input. The simulated results show that，compared with the conventional one-class support vector machine，the improved algorithm can be used to improve the detection accuracy which is up to 96.27% under different SNRs.

Key words: ship, wakedetection, featureextraction, adaptivethreshold, one-classsupportvectormachine

中图分类号:

TB566

王成，吴岩，杨廷飞. 利用改进单分类支持向量机提升舰船尾流目标的检测准确率[J]. 兵工学报, 2020, 41(9): 1887-1893.

WANG Cheng， WU Yan， YANG Tingfei. Improving the Detection Accuracy of Wake Targets by Improved One-class Support Vector Machine[J]. Acta Armamentarii, 2020, 41(9): 1887-1893.

参考文献

［1］ JIAO J J，HE Y，PAN X C，et al.Analysis on the factors that influence bubble coalescence in an acoustic field[J].Chinese Journal of Acoustics，2016，35(1):48-56.
[2］范雨喆，李海森，徐超，等.气泡线性振动时近海面气泡群的声散射[J].声学学报，2019，44(3):312-320.
FAN Y Z，LI H S，XU C，et al. Acoustic scattering of bubble groups in near-sea surface during linear vibration of bubbles[J].Acta Acoustics，2019，44(3):312-320. (in Chinese)
[3］田晶，白光富，江阳.一种基于偏振特征的激光尾流探测方法[J].兵工学报，2018，39(5):969-975.
TIAN J，BAI G F，JIANG Y.A laser wake detection method based on polarization characteristics[J].Acta Armamentarii，2018，39(5): 969-975.(in Chinese)
[4］ JIN L A，YAN X F，CHI W，et al. Numerical study method of the ship bubble wake based on LES and BPBE[J].Journal of Ship Mechanics，2015，19(1):28-34.
[5］张冲，袁志勇，张锐，等.舰船尾流气泡检测系统设计[J].兵器装备工程学报，2017，38(8):45-49.
ZHANG C，YUAN Z Y，ZHANG R，et al.Design of bubble detection system for ship wake[J].Journal of Ordnance Equipment Engineering，2017，38(8):45-49.(in Chinese)
[6］张群，王英民.尾流中气泡的主动声特性模型及有限元分析[J].压电与声光，2012，34(3):392-395，399.
ZHANG Q，WANG Y M .Active acoustic characteristic model and finite element analysis of bubbles in wake[J].Piezoelectric and Sound and Light，2012，34(3):392-395，399.(in Chinese)
[7］蔡平，朱代柱，惠俊英.水面船尾流的声散射实验研究[J].声学学报，2000，25(1):10-15.
CAI P，ZHU D Z，HUI J Y. Study on acoustic scattering of stern flow in surface[J].Acta Acoustics，2000，25(1):10-15.(in Chinese)
[8］杜敬林，马忠成，陈建辉，等.舰船尾流声散射和几何特征试验[J].哈尔滨工程大学学报，2010，31(7):909-914.
DU J L，MA Z C，CHEN J H，et al. Sound scattering and geome-trical characteristics test of ship wake[J].Journal of Harbin Engineering University，2010，31(7):909-914. (in Chinese)
[9］李婷.舰船尾流回波信号建模与仿真[J].计算机仿真，2011，28(6):26-29.
LI T.Modeling and simulation of ship wake echo signals[J].Computer Simulation，2011，28(6):26-29. (in Chinese)
[10］樊书宏，严冰，刘昆仑，等.舰船尾流侧向声检测方法[J].声学技术，2011，30 (6):474-479 .
FAN S H，YAN B，LIU K L，et al.A lateral sound detection method for ship wakes[J].Technical Acoustics，2011，30(6):474-479.(in Chinese)
[11］张有正.随机梯度遗传算法在船舶尾流目标识别中的应用[J].舰船科学技术，2017，39(6):25-27.
ZHANG Y Z.Application of stochastic gradient genetic algorithm in ship wake target recognition[J].Ship Science and Technology，2017，39(6):25-27. (in Chinese)
[12］王洪波.单分类支持向量机的学习方法研究[D].杭州：浙江大学，2012.
WANG H B.Research on learning method of one-class support vector machine[D].Hangzhou：Zhejiang University，2012.(in Chinese)
[13］ SYED FAWAD H.A novel robust kernel for classifying high-dimensional data using support vector machines[J].Expert Systems with Applications，2019，131：116-131.
[14］ WANG F，LIU S J，NI W C，et al.Imbalanced data classification algorithm with support vector machine kernel extensions[J].Evolutionary Intelligence，2019，12:341-347.
[15］戴卫国，李海涛，颜恒平，等.基于支持向量机改进算法的船舶类型识别研究[J].声学技术，2015，34(3):203-208.
DAI W G，LI H T，YAN H P，et al.Application of an improved support vector machine classification algorithm to underwater targets recognition[J].Technical Acoustics，2015，34(3):203-208.(in Chinese)
[16］ SCHOLKOPF B.Comparing support vector machines with Gaussian kernels to radial basis function classifiers[J].IEEE Transactions on Signal Processing，1997，45(11):2758-2765.
[17］ JEONG S，BAN S W，CHOI S，et al.Surface ship-wake detection using active sonar and one-class support vector[J]. IEEE Journal of Oceanic Engineering，2012，37(3):456-466.
[18］ LIN X G，JIAO Y Z，ZHAO D W.An improved Gaussian filter for dynamic positioning ships with colored noises and random measurements loss[J].IEEE Access，2018，6:6620-6629.

[1]	刘月, 刘铁林, 姜相争, 韩月明. 基于径向基函数和自组织映射的导弹控制系统健康状态评估[J]. 兵工学报, 2022, 43(4): 931-939.
[2]	王锴松，周国华，刘月林，王玉芬，刘胜道. 地磁模拟法测量舰船感应磁场的数值模拟[J]. 兵工学报, 2022, 43(3): 617-625.
[3]	郭成豹，胡松，王文井，殷琦琦. 利用磁传感器阵列磁场差值的舰船磁场反演建模方法[J]. 兵工学报, 2022, 43(1): 111-119.
[4]	李恒, 叶祖坤, 查文彬, 王禹林. 基于多传感器信息决策级融合的刀具磨损在线监测[J]. 兵工学报, 2021, 42(9): 2024-2031.
[5]	徐英，谷雨，彭冬亮，刘俊，陈华杰. 面向合成孔径雷达图像任意方向舰船检测的改进YOLOv3模型[J]. 兵工学报, 2021, 42(8): 1698-1707.
[6]	喻鹏，程锦房，张伽伟，姜润翔. 基于Rao检测器的舰船轴频电场滑动门限检测方法[J]. 兵工学报, 2021, 42(4): 827-834.
[7]	王逸南，张建伟，王治，姚熊亮，杨娜娜. 基于板厚补偿的921A钢与Q345钢靶板在截卵形弹体侵彻下的等效方法[J]. 兵工学报, 2021, 42(11): 2465-2475.
[8]	黄莎玲，朱鸿志，程祥利，谢玉斌. 自适应阈值侵彻引信层目标识别算法[J]. 兵工学报, 2020, 41(9): 1762-1771.
[9]	赵文春，姜润翔, 喻鹏，张伽伟. 基于轴频电场线谱特征的目标检测及识别[J]. 兵工学报, 2020, 41(6): 1165-1171.
[10]	陈浩，潘逊，肖大为. 模拟舰船磁场的两轴磁体设计研究[J]. 兵工学报, 2020, 41(6): 1172-1178.
[11]	孙强，姜润翔，喻鹏，程锦房. 轴频电场与静电场一体化腐蚀相关电场隐身方法研究[J]. 兵工学报, 2020, 41(4): 730-736.
[12]	余永维，彭西，杜柳青，陈天皓. 基于深度学习框架的装配机器人零件实时检测方法[J]. 兵工学报, 2020, 41(10): 2122-2130.
[13]	李茂，侯海量，李典，陈鹏宇，李永清，朱锡. 动态爆炸战斗部对舰船舱壁的破片载荷特性研究[J]. 兵工学报, 2019, 40(9): 1804-1818.
[14]	仝蕊，康建设，孙健，杨文，李宝晨. 基于局部特征尺度分解与复合谱分析的齿轮性能退化特征提取[J]. 兵工学报, 2019, 40(5): 1093-1102.
[15]	代健，李泽，郝新红，栗苹. 基于目标联合特征提取的脉冲多普勒引信抗干扰方法[J]. 兵工学报, 2019, 40(2): 225-233.