An air cavity method for increasing the underwater acoustic targets strength of corner reflector

In order to improve the underwater acoustic target strength of comer reflectors,according to the principle of acoustic impedance mismatch of the boundary layer,the method of using air cavity to increase the underwater acoustic target strength of corner reflectors is proposed.The acoustic reflection coefficients of underwater air layer and single layer metal sheet are calculated and compared.The results show that the reflection coefficient of single layer metal sheet is greatly affected by frequency and incidence angle,and the reflection coefficient of air layer in water is large and little affected by frequency and incidence angle.On this basis,a new kind of airfilled cavity corner reflector is designed.The acoustic scattering characteristics of underwater airfilled cavity comer reflector are calculated cumulatively,and the results are compared with the monolayer metal sheet corner reflector.The simulation results show that the acoustic reflection effect of the airfilled cavity corner reflector is better.In order to verify the correctness of the method,the test was carried out in the silencing tank.The experimental results show that the simulation results are in good agreement with the experimental results,and the airfilled cavity can improve on acoustic reflection performance of the underwater corner reflector.

Underwater Noise Target Recognition Based on Sparse Adversarial Co-Training Model with Vertical Line Array

The automatic identification of underwater noncooperative targets without label records remains an arduous task considering the marine noise interference and the shortage of labeled samples.In particular,the data-driven mechanism of deep learning cannot identify false samples,aggravating the difficulty in noncooperative underwater target recognition.A semi-supervised ensemble framework based on vertical line array fusion and the sparse adversarial co-training algorithm is proposed to identify noncooperative targets effectively.The sound field cross-correlation compression(SCC)feature is developed to reduce noise and computational redundancy.Starting from an incomplete dataset,a joint adversarial autoencoder is constructed to extract the sparse features with source depth sensitivity,aiming to discover the unknown underwater targets.The adversarial prediction label is converted to initialize the joint co-forest,whose evaluation function is optimized by introducing adaptive confidence.The experiments prove the strong denoising performance,low mean square error,and high separability of SCC features.Compared with several state-of-the-art approaches,the numerical results illustrate the superiorities of the proposed method due to feature compression,secondary recognition,and decision fusion.

Bark-Wavelet Analysis and Hilbert -Huang Transform for Underwater Target Recognition

Recognizing the underwater targets by the radiated noise information is one of the most significant subjects in the area of underwater acoustics. Based on the theory of auditory perception, a novel recognition approach which consists of the algorithms of Bark-wavelet analysis, Hilbert-Huang transform and support vector machine is proposed. The performance of the proposed method is validated by comparing with traditional method and evaluated by the recognition experiments for SNRs of 0 dB, 5 dB, 10 dB, 15 dB and 20 dB.The results show that the average recognition rate of the method is above 88% and can be increased by 0.75 % to 6.25% under various SNR conditions compared to the baseline system.

An Efficient Acoustic Scattering Model Based on Target Surface Statistical Descriptors for Synthetic Aperture Sonar Systems

Acoustic scattering as the perturbation of an incident acoustic field from an arbitrary object is a critical part of the targetrecognition process in synthetic aperture sonar(SAS)systems.The complexity of scattering models strongly depends on the size and structure of the scattered surface.In accurate scattering models including numerical models,the computational cost significantly increases with the object complexity.In this paper,an efficient model is proposed to calculate the acoustic scattering from underwater objects with less computational cost and time compared with numerical models,especially in 3D space.The proposed model,called texture element method(TEM),uses statistical and structural information of the target surface texture by employing non-uniform elements described with local binary pattern(LBP)descriptors by solving the Helmholtz integral equation.The proposed model is compared with two other well-known models,one numerical and other analytical,and the results show excellent agreement between them while the proposed model requires fewer elements.This demonstrates the ability of the proposed model to work with arbitrary targets in different SAS systems with better computational time and cost,enabling the proposed model to be applied in real environment.

Modeling and simulation of torpedo acoustic homing trajectory with multiple targets

The characteristics of a torpedo＇s acoustic homing trajectory with multiple targets were studied. The differential equations of torpedo motion were presented based on hydrodynamics. The Fourth order Runge-Kutta method was used to solve these equations. Derived from sonar equations and Snell＇ s law, a simple virtual underwater acoustic environment was established for simulating the torpedo homing process. The Newton iteration method was used to calculate homing range and ray tracing was approximated by pieccwise line, which takes into consideration distortions cause by temperature, pressure, and salinity in a given sea area. The influence of some acoustic warfare equipment disturb the torpedo homing process in certain circumstances, including decoys and jammers, was alsotaken into account in simulations. Relative target identification logic and homing control laws were presented. Equal consideration during research was given to the requirements of rcal-timeactivity as well as accuracy. Finally, a practical torpedo homing trajectory simulation program was developed and applied to certain projects.

基于CNN和DCGAN的小样本船舶辐射噪声识别方法

文中建立一种基于卷积神经网络(convolutional neural networks,CNN)和深度卷积生成对抗网络(deep convolutional generative adversarial networks,DCGAN)的船舶目标识别方法.通过采集的船舶辐射噪声数据,以梅尔频谱(Mel spectrogram)作为网络的输入特征,使用DCGAN网络对频谱变换后的样本进行扩充,利用微调的VGG16(visual geometry group)网络实现船舶目标分类,实现了网络收敛速度的提升和训练时间的减少.结果表明:采用所提方法可以生成较高质量的频谱样本,提高船舶辐射噪声识别的准确率.

基于面积加权GWT-GFT的水声目标识别

由于海洋环境的复杂性,水声目标的识别具有很大的挑战性。为解决这类复杂环境下特征提取的问题,提出了一种基于面积加权的图小波变换-图傅里叶变换(GWT-GFT)的分析方法。在完成数据预处理后,为了能够凸显顶点之间的关系,提出了一种新的基于顶点三角形面积的加权方法来构建图信号;构建好的图信号通过GWT分解为多尺度图分量;然后,利用GFT将这些分量从图域变换到特征值谱域进行分析;在此基础上,提取各分量特征值谱的特征;最后,利用基于高斯核函数的支持向量机(SVM)对获取的特征向量进行分类。基于水声信号ShipsEar数据库,采用5折交叉验证方法进行验证。与现有的其它方法相比,所提的模型以36个特征在376656个样本上取得了97.22%的准确率,证明了该分析方法的有效性和鲁棒性。

On fast estimation of direction of arrival for underwater acoustic target based on sparse Bayesian learning

The Direction of Arrival （DOA） estimation methods for underwater acoustic target using Temporally Multiple Sparse Bayesian Learning （TMSBL） as the reconstructing algorithm have the disadvantage of slow computing speed. To solve this problem, a fast underwater acoustic target direction of arrival estimation was proposed. Analyzing the model characteristics of block-sparse Bayesian learning framework for DOA estimation, an algorithm was proposed to obtain the value of core hyper-parameter through MacKay＇s fixed-point method to estimate the DOA. By this process, it will spend less time for computation and provide more superior recovery performance than TMSBL algorithm. Simulation results verified the feasibility and effectiveness of the proposed algorithm.

Acoustic localization scheme and accuracy analysis for underwater vertical moving target using seabed stations

To measure the trajectory of an underwater vertical moving target(UVMT) in transient motion with high accuracy and high frame rate,an acoustic localization model using seabed stations with an acoustic beacon was presented.A solution algorithm based on the Gauss-Newton method was derived,which was shown to satisfy the local linear convergence.Accuracy analysis of the numerical simulation indicated that the station location,sound velocity,and signal time delay estimation errors were propagated to location parameters through measurement ranges,and the main affecting factors included the station geometry,target relative location,and acoustic conditions.Vertical accuracy was improved using a supplemental surface station coupled with the seabed stations.Detailed characteristics were indicated by accuracy distribution from the full test sea area.A 14-station array composed of 13 seabed stations and 1 surface station in a test sea of 1 km x 1 km and 60 m in depth demonstrated that the average root mean square errors(RMSEs) in the x,y,and z directions were 0.30,1.47,and0.34 m,respectively,in the vertical range of 35-60 m.This work provided a technical approach for UVMT localization,which would be useful for designing related measurement systems.

Extraction and application of the low dimensional dynamical component from underwater acoustic target radiating noise

Signal processing in phase space based on nonlinear dynamics theory is a new method for underwater acoustic signal processing. One key problem when analyzing actual acoustic signal in phase space is how to reduce the noise and lower the embedding dimen- sion. In this paper, local-geometric-projection method is applied to obtain fow dimensional element from various target radiating noise and the derived phase portraits show obviously low dimensional attractors. Furthermore, attractor dimension and cross prediction error are used for classification. It concludes that combining these features representing the geometric and dynamical properties respectively shows effects in target classification.

加权有向关联网络构建与表征的水中目标远距离检测

水中目标的远距离检测是海洋防御体系的关键技术之一,对国防及民用领域均具有十分重要的作用。然而,目前尚缺乏行之有效的水中目标远距离检测方法,特别是目标先验信息未知的情况下变的愈加困难。为解决这一问题,提出一种新的方法—加权有向关联网络。通过矢量声信号到加权有向关联网络的映射,将信号检测问题转化为网络拓扑的表征,并通过对网络拓扑的特性分析及特征提取,实现无目标先验信息下的水中目标远距离检测。并通过仿真与实测数据对所提出的方法进行验证。研究结果表明:与现有的窄带互谱检测、冒泡熵等方法相比,所提方法能够检测到更低信噪比的水中目标,实现了无需目标先验信息的水中目标远距离检测;该方法的应用具有一定的实际意义和应用前景,可以为海洋防御和民用领域的水下目标检测提供有效的技术支撑。

基于MEMS前沿传感技术的水声专业教学实践研究

针对信息技术应用模式快速发展对创新型人才培养的迫切需求,围绕着水声工程创新发展要求,提出将压电MEMS芯片传感前沿技术与学生毕业设计题目关联,开展毕业设计实践,突破原有毕设题目独立设置模式,并通过与科研院所联合开展学生实践活动,使学生在较短的时间内掌握压电MEMS芯片前沿技术背景、关键技术以及与水声工程应用技术内在支撑关系,为专业领域新型人才培养模式探索提供了经验。

基于深度可分离卷积神经网络的水声目标分类研究及FPGA实现

针对传统声纳处理器算力受限,能效比低,难以支撑水声目标识别实时推理的问题,本文基于异构SoC平台设计了面向被动声纳水下目标实时计算处理系统。该系统具有较低资源开销和较小分类精度损失等优点,是一种低时延、高能效比的硬件加速器解决方案。本文以MobileNetV1网络模型为基础并对其进行结构优化,在现场可编程门阵列(Field programmable gate array,FPGA)上通过并行流水线的加速结构实现它的前向推理过程,并对其权值参数进行二值化的处理,以达到减少存储量和计算量的同时加快其推理速度的目的。同时,根据在输入通道维度以及输出图像高度上分块并行的优化思想,设计了深度可分离卷积的流水优化策略,采用并行流水的结构极大减少了前向推理的时间。实验表明,在利用出海实际采集得到的水声数据集上,本文实现的系统识别精度为88.5%,在的分辨率的图像上,时间延迟达到4.23 ms。对比CPU速度提升了70.68倍,是GPU速度的68%。能效比分别为CPU的0.08%,GPU的2.12%。本文为神经网络在硬件资源有限以及功耗存在限制的轻量型移动端或者边缘设备上的应用与部署,以及对促进融合水下勘探网络的建设和水下信息的快速获取提供了设计思路。

一种基于冲激脉冲回波检测的主动目标定位技术

水下等离子体声源为一种强度高、宽带可控的脉冲声源,其声源级高于传统换能器,常被用作干扰远程目标的软杀伤性武器。当其产生的冲击波无法对目标实施有效的毁伤和干扰等软杀伤打击时,本文研究利用脉冲回波检测来对目标进行主动定位,以提升其运用效能。研究水下等离子体声源脉冲声波的有关特性,并对海底混响和海洋环境噪声与主动定位的关系进行相关仿真研究。建立以脉冲声波为主动声信号的目标主动定位处理模型,以舷侧11元阵和舷侧三子阵为接收水听器阵列,研究基于等离子体声源的主动目标定位技术,仿真验证了方法的有效性。

Local-linear-prediction analysis for underwater acoustic target radiated noise

Local-linear-prediction in phase space is performed for the underwater acoustic target radiated noise. Relation curve of average prediction error versus neighboring points' number is calculated. The result is used in judging the nonlinearity of radiated noise time series, and obtaining the appropriate form and coefficients of predicting model. The line and continuous spectral component are predicted respectively. Choice of some model parameters minimizing the prediction error is also discussed.

海底掩埋目标声探测与识别关键技术进展

为了准确呈现海底掩埋目标声探测与识别研究存在的问题与技术难点,推动该领域相关技术的发展,本文主要从探测系统及信号处理技术两方面出发进行研究。首先分析了基于水声探测海底掩埋目标所面临的难点,介绍了国内外掩埋目标声探测与识别系统,然后论述了抗混响、特征提取、目标检测和目标识别等关键技术环节的主要研究进展,最后总结了掩埋目标探测与识别面临的问题和挑战,并对未来发展方向进行了展望。

基于Kriging代理模型的水下目标模型几何参数识别方法

[目的]水下目标参数识别可为目标分类识别提供依据,为此,提出一种基于Kriging代理模型的水下目标参数识别方法。[方法]首先,对敷设声学覆盖层的水下目标模型在螺旋桨和主辅机激励情况下的结构表面低频振动声辐射与声辐射灵敏度进行分析;然后,基于分析结果建立低频声辐射功率代理模型,并基于该代理模型构造由低频声辐射响应特征和目标参数组成的样本空间;最后,基于所构建的样本空间,建立目标参数识别代理模型并选取测试点进行模型验证。[结果]结果显示,测试样本的实际目标参数值与所构建代理模型的目标参数预测值吻合良好;利用有限元法和边界元方法可以实现考虑阻尼材料频变特性的黏弹性阻尼结构的低频声辐射分析,并能解决商业软件无法大批量处理振动结果文件的问题;影响水下目标模型低频振动声辐射的主要目标参数为目标长度、最大半径、基层壳厚度和声学覆盖层厚度。[结论]基于Kriging代理模型的水下目标参数识别方法可以通过声辐射线谱特征准确预测水下目标模型的主要目标参数值。

联合线谱增强与深度神经网络的水声目标识别

为强化水声目标特征,提高使用深度神经网络识别水声目标的准确率,提出了一种联合线谱增强与深度神经网络的水声目标识别方法。该方法采用窄带信息增强,将自适应线谱增强滤波器与VGGish神经网络级联,水声信号经过线谱增强后输入网络提取深度特征,之后使用分类器分类。使用实测水声数据集进行测试,对网络提取的水声数据的深度特征集进行主成分分析并降维,使高维深度特征可视化,结果表明线谱增强后得到的深度特征集的紧致性明显提高。该方法在测试数据集上能够实现94.83%的识别准确率,与未进行线谱增强的情况相比提升了5.48%,同时在低信噪比情况下稳定性更好。

一种基于VGGish神经网络的水声目标识别方法

水声目标智能识别是水声装备智能化的重要组成部分,深度学习则是实现水声目标智能识别的重要技术手段之一。当前水声目标智能识别经常面临数据集较小带来的训练样本量不足的情况,针对小数据集识别中存在的因过拟合导致模型泛化能力不足,以及输入的水声信号二维谱图样式不统一的问题,文章提出了一种基于VGGish神经网络模型的水声目标识别方法。该方法以VGGish网络作为特征提取器,并在VGGish网络前部加入了信号预处理模块,同时设计了一种基于传统机器学习算法的联合分类器,通过以上措施解决了过拟合问题和二维谱图样式不统一问题。实验结果显示,该方法应用在ShipsEar数据集上得到了94.397%的识别准确率,高于传统预训练-微调法得到的最高90.977%的准确率,并且在相同条件下该方法的模型训练耗时仅为传统预训练-微调方法的0.5%左右,有效提高了识别准确率和模型训练速度。

多目标水声信号的稀疏重构反卷积测向算法

针对浅海复杂定位环境下信噪比低、多信源目标方位估计分辨能力低的问题,文章提出了多目标水声信号的离格稀疏贝叶斯学习重构反卷积测向算法。首先,该算法利用维纳滤波反卷积算法对阵元接收的信号进行“去噪”处理,然后对信号数据进行奇异值分解,从而降低噪声和信号重构过程的计算量;再建立离格稀疏信号模型,通过贝叶斯学习算法得到最大后验概率;最后求出多个目标信源的波达方向估计值。文章所提算法通过使用维纳滤波反卷积超分辨算法,获得了更高的方位估计的分辨率,提高了对多个目标的检测性能。仿真分析和海试实验数据结果表明,与MUSIC算法和OGSBI算法相比,该方法在信噪比大于-8 dB时,方位估计的均方根误差在1°以内,并在多目标定位精度、算法鲁棒性以及运行速度上均有更优的性能,为水下多目标波达方向估计提供了参考。