水中运动目标动态线谱增强算法研究

通常的基于短时自相关的自适应线谱增强器(SABALSE)主要缺点是:输入信噪比低时,抑制高斯噪声性能差。为了最大限度地克服SABALSE的缺点,我们充分利用高阶累积量抑制高斯噪声的性能和高阶累积量的不同更新算法,提出了基于四阶累积量不同切片的自适应动态线谱增强新算法,并对其原理、结构进行了剖析。用实测鱼雷线谱数据,对鱼雷与水听器处于不同方位时,水听器接收的线谱进行了动态仿真。结果表明:基于四阶累积量非对角切片的自适应动态线谱增强(NDSCBADLSE)算法抑制高斯噪声、增强动态线谱的能力强于基于对角切片的自适应动态线谱增强(DSCBADLSE)算法,且均强于SABALSE算法。因此,本文的算法可用于提高水下探测系统和水下武器系统对微弱信号的检测能力。     

用于拖曳式线列阵的一种新的线谱增强系统

拖曳式线列阵是近来声呐信号处理领域中非常活跃的课题.自适应线谱增强器有可能用于拖曳式线列阵的信号处理系统.在利用LMS迭代算法的自适应线谱增强器中,由于存在迭代噪声,系统增益与输入信噪比有关.如果不抑制拖曳平台的噪声,线谱增强系统的效用就无法发挥出来,文中提出用自适应噪声抵消系统和自适应线谱增强器级联的概念.首先把拖曳平台的噪声作为干扰加以抑制,再对输出信号进行线谱增强,从而可以提取微弱信号的线谱分量.     

基于分段信号相关累加的变速度多站联合直接定位方法

在低信噪比条件下,基于时延和多普勒频移的直接定位算法在解决宽带信号源定位时精度较差.针对此问题,提出了一种基于分段信号相关累加的变速度多站联合直接定位算法,并给出了其克拉美罗下界.该算法利用多个变速度的观测站对信号进行接收,然后将同一观测站接收的目标信号分割成多段不重叠的短时信号,采用最大似然估计器,联合各段信号的时延、多普勒频移信息对目标进行直接定位.算法充分利用了观测信号包含的定位信息,并利用观测站速度的变化增加了目标位置信息,解决了分段信号联合估计带来的定位模糊问题,使定位精度进一步提高,增加了算法的实用性.仿真实验表明,较之传统直接定位算法,本文算法定位精度更高,尤其在低信噪比条件下更能逼近克拉美罗界.     

水下目标辐射噪声中单频信号分量的检测:理论分析

在理想情况下高斯噪声背景下高斯信号的最佳检测问题已受到很多关注,它是被动声呐采用能量检测的理论基础之一.但是理论与实验都已证明,舰船辐射噪声中存在着单频信号分量.在小信噪比下,单频信号的检测必须采用有别于宽带能量检测的方法.本文讨论在小信噪比下检测单频信号分量的几种不同的理论方法,包括自相关检测、FFT分析、自适应线谱增强等,计算了它们的增益.指出分段FFT检测具有明显的优势,对可能出现的单频信号分量的频率漂移具有较好的宽容性.     

几种自适应IIR陷波器的复数算法及其应用

在自适应IIR陷波器实数算法的基础上,本文提出了三种基于梯度的自适应IIR陷波器的复数算法,并进行了仿真比较。结果表明,这三种复数算法均可检测复信号,并估计信号频率,其中"改进的"简化格型IIR陷波器复数算法收敛速度快、低信噪比下稳定性能好。湖上试验表明该算法不仅实时性好,而且在低信噪比下的检测效果令人满意。     

水下线谱噪声源识别的波束域时频分析方法研究

从联合空时频三维信息从发, 提出了波束域时频分析识别水下运动航行器低频线谱噪声源位置的方法. 首先, 利用小孔径圆环阵的超指向性波束形成, 将各线谱噪声源匀速通过正横位置附近时产生的多普勒信号在时域上分离. 其次, 分别使用伪Wigner-Ville分布和调频小波变换两种时频分析方法对波束输出的信号进行处理, 得到各噪声源信号的时频图像. 最后, 转换时间坐标到空间并参考配置信标, 即可识别低频线谱噪声源在水下航行器上的位置. 该方法解决了阵列识别水下低频噪声源的孔径受限问题, 同时对处理同频相干噪声源也适用. 仿真试验结果表明: 两种波束域时频分析方法都能较精确地识别低频线谱噪声源的位置; 在测量系统信息的配合下, 波束域调频小波变换的识别效果更优.     

利用深层卷积网络自适应增强学习的水声目标线谱提取方法

提出了一种使用自适应增强学习的深层卷积神经网络方法对水声目标线谱进行提取。该方法利用构造的多尺度ConvNeXt算子建立滑动窗深层卷积神经网络模型(SwDCNN),设计涵盖损失函数、学习率更新和模型迭代优化的自适应增强学习准则并用于模型训练。仿真和海试数据验证结果表明,所提方法有以下优点:(1)卷积算子和模型结构参数按线谱提取需求配置,可以增强LOFAR谱图特征高性能多分辨力挖掘能力;(2)大规模数据下的模型训练可实现渐进式精确拟合,有助于提升模型收敛效果;(3)模型可有效提取低信噪比、中断、弯曲漂移、粗细不均、邻近成簇、密集分布等复杂情况下的线谱,在查全率、查准率、虚警率、线谱位置精度(LLA)和线谱幅值精度(LAA)等指标上均优于文中其他深度神经网络方法;(4)和传统及其他文中所用的深度神经网络方法相比,线谱最小可检测信噪比分别降低超过5 dB和2 dB,实际复杂场景线谱提取能力更强,综合效果更好。     

相干多普勒测风激光雷达低信噪比区域回波信号的估计方法

相干多普勒测风激光雷达通常会采用周期图最大值法（PM）提取不同距离门信号的多普勒频移（对应风速）信息。由于噪声和相干效率的影响,个别距离门信号会出现信噪比（SNR）突然降低的情况,从而导致系统的探测概率降低,影响系统整体的探测性能。为了解决个别距离门信号多普勒频移的错误估计问题,提出了一种新的非线性自适应多普勒频移估计方法。该方法利用风速的连续性,标定错误距离门,并自适应地利用强信噪比区域的多普勒频移统计数据来弥补信噪比变差而出现的估计错误。分别利用了仿真模型和一套1.54μm全光纤相干激光雷达系统获得了风场测量数据,对比了使用该技术前后反演得到的风速趋势,证明该方法能够有效地解决上述问题。     

空时多普勒频移域运动小目标的抗干扰探测方法

有源声呐使用多普勒敏感信号可在多普勒频移域分离杂波、混响等干扰和运动目标,但低信干比时强干扰的时间旁瓣泄露和多普勒旁瓣泄露会淹没弱目标.针对该问题,提出了一种基于自适应最小均方(LMS)算法和宽带模糊函数(WAF)的运动小目标抗干扰探测方法.首先在空间多普勒频移域上利用基于LMS的自适应陷波器抑制干扰,然后在快时间多普...     

基于二级自适应滤波的水下目标动态谱线增强算法研究

为了从非高斯噪声环境中有效提取水下目标辐射的线谱信号,以提高水下探测系统检测水下目标的能力,提出了基于二级自适应滤波的水下目标动态谱线增强算法,该算法用基于高阶累积量能量幂函数变步长极性迭代的自适应谱线增强器作为第一级,用传统的自适应谱线增强器(ALE)作为第二级,实行串级联接。用输入信号的峭度定义了峭度信噪比,并用此分析了该算法的性能。用水下某目标辐射线谱的实测数据,对该算法的性能进行了仿真研究。结果表明:当环境噪声为瑞利噪声或混合噪声(包含均匀分布、瑞利分布和拉普拉斯分布的噪声成分)时,该算法与ALE算法相比,有良好的抑制抑高斯噪声或非高斯噪声、提高信噪比和跟踪时变信号的性能。     

稀疏驱动自适应线谱增强的水下目标谱熵检测

针对无人平台在水下复杂环境中的线谱弱目标自主检测问题,提出了一种采用稀疏驱动自适应线谱增强(ALE)为前处理的监督学习目标检测方法。该方法在ALE代价函数中引入稀疏性l_p范数,并将稀疏正则化推广到01/2稀疏驱动ALE比常规ALE的处理增益高11.5 dB。利用水下无人平台海上拉距试验的数据对算法性能进行验证,在宽带强干扰影响下,该方法可有效检测远距离声源,虚警率为3.5%时,检测率达95.8%,有效提高了对线谱弱目标的检测概率,具有较强的环境适应性。      

A novel algorithm for underwater moving-target dynamic line enhancement

When we use a traditional adaptive line enhancer (ALE) algorithm to detect an underwater moving target, there are two disadvantages: the ability to suppress colored Gaussian noise is low and the lower the SNR is, the worse the performance of the ALE algorithm. In order to greatly overcome these disadvantages, we take full advantage of the capability of higher order cumulants to alleviate the effect of colored Gaussian noise and develop a fourth order cumulant non-diagonal slice-based adaptive dynamic line enhancer (FOCNDSBADLE) algorithm and fourth order cumulant diagonal slice-based adaptive dynamic line enhancer (FOCDSBADLE) algorithm. The adaptive filtering coefficients of these algorithms are indirectly updated by the instantaneous fourth order cumulant slices. It is shown that these slices are comprised of noiseless sinusoids if the input signals are comprised of sinusoids corrupted by Gaussian noise. Therefore these algorithms are fit to handle highly colored Gaussian noise. Simulation tests are carried out using the measured data radiated by the underwater moving target. Simulation results have shown that the FOCNDSBADLE algorithm and FOCDSBADLE algorithm outperform the ALE algorithm and that the FOCNDSBADLE algorithm outperforms the FOCDSBADLE algorithm in the case of Gaussian noise.     

Wind sensing in an atmospheric boundary layer by means of micropulse coherent Doppler lidars

An algorithm is developed and computer simulation of wind sensing by means of micropulse coherent Doppler lidars (CDLs) in the atmospheric boundary layer is conducted for low values of the signalto- noise (SNR) ratio. The accuracy of lidar wind measurements is studied numerically for parameters of micropulse Stream Line CDLs. Optimal parameters of the measurements and processing data obtained at low SNR, which allow reconstructing vertical profiles of the wind velocity vector with required accuracy within an entire atmospheric boundary layer, are determined.     

Active vibration control using optimized modified acceleration feedback with Adaptive Line Enhancer for frequency tracking

Modified acceleration feedback (MAF) control, an active vibration control method that uses collocated piezoelectric actuators and accelerometer is developed and its gains optimized using an optimal controller. The control system consists of two main parts: (1) frequency adaptation that uses Adaptive Line Enhancer (ALE) and (2) an optimized controller. Frequency adaptation method tracks the frequency of vibrations using ALE. The obtained frequency is then fed to MAF compensators. This provides a unique feature for MAF, by extending its domain of capabilities from controlling a certain mode of vibrations to any excited mode. The optimized MAF controller can provide optimal sets of gains for a wide range of frequencies, based on the characteristics of the system. The experimental results show that the frequency tracking method works quite well and fast enough to be used in a real-time controller. ALE parameters are numerically and experimentally investigated and tuned for optimized frequency tracking. The results also indicate that the MAF can provide significant vibration reduction using the optimized controller. The control power varies for vibration suppression at different resonance frequencies; however, it is always optimized.     

核磁共振显微术的DESIRE效应模拟(英文)

要在合理时间內进行微米等级解析度的扫描时,核磁共振波谱的信噪比一直是一个很大的限制. DESIRE效应利用扩散现象来增强信号强度及解析度的概念被提出后,信噪比在一维的情况下大约可以增加一个数量级.在先期的研究中,模拟外加磁场的效应通常是使用叠代数值计算,而扩散的效应则是利用有限微分法解Bloch-Torrey e-quation ,这样的模拟方法通常很耗时.该研究中,分别使用Shinnar-LeRoux演算法及基于旋积的扩散模拟方法来加速计算出外加磁场及扩散对于信号的影响,因此可以节省大量计算时间以进行复杂的计算.利用一维模拟结果预测在不同参数实验中解析度的改变和信号增加值,并找出最佳的参数值.而由模拟结果指出,Sinc3脉冲波形可以得到最佳的信号增加量及不错的解析度.     

ALE计算中基于声速分布的网格优化技术

基于声速分布,提出一种拉氏流体力学计算中大变形网格优化的数值技术.该方法不但可以优化网格的几何形状且可以提高拉氏流体计算的时间步长.介绍基于声速分布的网格松弛泛函、修正梯度流方程的推导、离散和求解方法,启动/终止网格优化过程的条件,及基于这种网格优化方法的ALE算法.给出Rayleigh-Taylor不稳定性问题等数值算例,用以证明该方法的有效性.     

Background noise cancellation of manatee vocalizations using an adaptive line enhancer

The West Indian manatee (Trichechus manatus latirostris) has become an endangered species partly because of an increase in the number of collisions with boats. A device to alert boaters of the presence of manatees is desired. Previous research has shown that background noise limits the manatee vocalization detection range (which is critical for practical implementation). By improving the signal-to-noise ratio of the measured manatee vocalization signal, it is possible to extend the detection range. The finite impulse response (FIR) structure of the adaptive line enhancer (ALE) can detect and track narrow-band signals buried in broadband noise. In this paper, a constrained infinite impulse response (IIR) ALE, called a feedback ALE (FALE), is implemented to reduce the background noise. In addition, a bandpass filter is used as a baseline for comparison. A library consisting of 100 manatee calls spanning ten different signal categories is used to evaluate the performance of the bandpass filter, FIR-ALE, and FALE. The results show that the FALE is capable of reducing background noise by about 6.0 and 21.4 dB better than that of the FIR-ALE and bandpass filter, respectively, when the signal-to-noise ratio (SNR) of the original manatee call is -5 dB.     

Acceleration target detection based on LFM radar

In radar systems, the echo signal caused by an accelerated target can be similarly considered as linear frequency modulation (LFM) signal. In high signal-to-noise ratio (SNR), discrete polynomial-phase transform (DPT) algorithm can be used to detect the echo signal, as it has low computation complexity and high real-time performance. However, in low SNR, the DPT algorithm has a large mean square error of the rate of frequency modulation and a low detection probability. In order to detect LFM signal in low SNR, this paper proposes a detection method, segment discrete polynomial-phase transform (SDPT), which means, at first, dividing the whole echo pulses into several segments with same duration in time domain, and then, using coherent accumulation method of DFT to segments, at last, processing this signal with DPT in intra-segment. In the case of a large number of segments, the SDPT can improve the output SNR. In addition, in a certain SNR, to the target signal with big sampling interval, large acceleration and less segments, this paper proposes an algorithm to detect the LFM signal generated from the combination of an improved DPT (IDPT) and fractional Fourier transform (FRFT). The output SNR of this algorithm is connected with the length of time delay. In the simulation, when the length of the time delay is 0.2 N, the output SNR is 2.5 dB more than that which results from directly using DPT. Finally, the detection performance and algorithm complexity of the proposed algorithm were analyzed, and the simulated and measured data verify the effectiveness of the algorithm.     

低信噪比下的二维联合线性布雷格曼迭代快速超分辨成像算法

针对实际逆合成孔径雷达(ISAR)成像时带宽有限、方位孔径稀疏的小角度回波数据条件下,常规算法的成像分辨率不高等问题,基于压缩感知理论,提出了一种低信噪比条件下的二维联合布雷格曼迭代快速ISAR超分辨成像算法.首先,将雷达回波构建为距离频域-方位多普勒域的二维稀疏表示模型,在此基础上,将二维超分辨成像问题转换为二维联合压缩感知的稀疏重构问题;其次,为了避免重构时向量化操作带来的复杂度,提出了二维联合布雷格曼迭代算法,为实现快速重构,将加权残量迭代、估计停滞步长与感知矩阵条件数优化三种加快收敛速度的思想相结合,既利用了布雷格曼迭代在低信噪比条件下的重构能力又能保证快速成像.最后仿真实验结果表明在欠采样和低信噪比条件下本文算法能够缩短成像时间,且具备更好的噪声鲁棒性.     

基于多结构元素形态滤波与自适应阈值分割相结合的红外弱小目标检测

针对低信噪比灰度图像中弱小目标检测的难题,分析了红外弱小目标成像的特点,提出了基于多结构元素形态滤波与自适应阈值分割相结合的目标检测算法.利用目标运动的连续性、规律性和噪音产生的随机性,结合数学形态学结构元素的特点,研究了一种多结构元素形态滤波的管道滤波方法,通过流水线管道检测目标运动轨迹.实验结果表明,该算法应用于复杂背景下低信噪比的红外弱小目标图像能够得到较理想的结果,并且目标检测概率高,速度快,虚警率低.