【作者】 李关防；

【导师】 惠俊英；

【作者基本信息】 哈尔滨工程大学 ， 水声工程， 2009， 博士

【摘要】 空中目标因其高效的搜索性能和机动性强的特点,在海空对抗中占据优势地位,对水下平台的生存构成威胁。本文以海空对抗中被动声探测技术为研究对象,对海空对抗声纳涉及的若干关键技术进行了研究,旨在为发展、实现海空对抗声纳进行有益的探索。希尔伯特黄变换是一种非线性信号处理理论,通过仿真研究了经验模态分解的突出信号局部特征与多分量信号分离特性,总结提出了不同用途下固有模态函数(IMF-Intrinsic Mode Functions)选取的原则,可用于信号去噪、互相关峰检测和多目标分辨等多种场合；回顾了平均声强器、互谱、能量加权直方图与解析声强流等常见的基于单矢量传感器的方位估计理论；将接收阵元组成阵,介绍了阵列输出模型及均匀直线阵的自然指向性,突出了阵增益及矢量的优势。空中目标辐射噪声频谱包括线谱和连续谱两部分,相比于宽带连续谱,线谱是其主要噪声源；从线谱和连续谱特征提取出发,提出了一种基于经验模态分解(EMD-Empirical Mode Decomposition)的线谱和连续谱分离算法。声波经过空-水界面后,部分声能透射到水中,激发的水下声场亦具有明显的线谱特征。在低频段,声波的声压、振速存在相位差,声能流在有功分量和无功分量间发生“串漏”,传统测向算法是有偏的。本文从试验的角度探索了基于水下声场的空中目标测向方法,分析讨论了将辐射噪声的线谱集中频段、多普勒频偏相对于原频率的比率和线谱的方位(变化趋势)信息作为特征量的空中目标的检测与识别方案；同时,目标的大致方位信息有利于水下平台对其的规避和有效的攻击,有望改善其的对空劣势。物体的入水声、水下目标启动声等是对目标探测非常重要的一类瞬态信号,从物体入水声形成机理上研究了其波形特性,为物体入水声的识别和击水声的源级预报提供理论基础。结合EMD的特性和平均声强器适合连续谱明显的单目标方位估计的优势,提出了模态声强器算法,利用模态声强器(MAIA-Mode Acoustic Intensity Averager method)算法可有效实现平台干扰下,物体入水声、目标启动声的检测和测向,能实现对瞬态信号信号检测。矢量阵最小方差无畸变响应波束形成算法可实现全空间内信号的空间谱估计,但该算法对弱目标的检测能力较差。基于此,结合IMF窄带特性与MVDR(Minimum Variance Distortionless Response)的窄带信号要求,提出了矢量阵模态域MVDR波束形成算法,将分解得到的声压、振速对应阶IMF利用MVDR波束形成估计其方位谱,并将中心频率的概念应用于IMF,综合利用声压、振速对应阶IMF求解中心频率,以此中心频率作为MMVDR (Mode domain Minimum Variance Distortionless Response)算法的中心频率。海试结果表明：MMVDR算法在保持强干扰空间谱能量不变的情况下,有效的增大了弱目标所在方位空间谱的能量,可实现强干扰下弱目标的检测,增加弱目标的有效探测距离。指向最小方差波束形成是一种高分辨波束形成算法,具有低旁瓣和主瓣更窄的优势。通过分析算法的计算量的影响因素,指出STMV(Steered Minimum Variance algorithm)算法计算量近似正比于阵元数M的三次方。当阵阵元数M较大时,STMV波束形成算法需很大计算开销,限制了其应用。基于此,结合α滤波器的积分器特性和矩阵求逆迭代公式,提出了一种STMV快速迭代算法,其计算量近似正比于M2；仿真验证了该算法在计算量降低的同时,保持了STMV算法高方位分辨力和弱目标检测的优点；应用于48元阵海试数据处理,计算量降低了约24倍,有利于算法的实时实现；同时可在多个目标信噪比相差较大和多个目标方位比较接近时,有效地检测到弱目标。更多还原

【Abstract】 The air targets such as helicopters are dominant in the aeronaval antagonism system because of its efficient hunting performance and agile maneuverability, which is a great threat to the existence of submarines.The subjects investigated in this dissertation are some pivotal passive detecting techniques related to the aeronaval antagonism system aimed to do some helpful research in the development and realization of aeronaval-antagonism SONAR system.Hilbert-Huang Transform (HHT) is a kind of nonlinear signal processing theory. Simulation results have approved that empirical mode decomposition (EMD) can emphasize the local instantaneous characters of original signals and decomposes the multi-component signals into mono-component signals. Aiming at different actual applications, the selection principles of intrinsic mode functions (IMF) are proposed, which could be used in denoising, cross-correlation peak detection and multiple-target recognition. The common single vector sensor azimuth estimation theories are reviewed such as acoustic intensity averager, cross-spectrum method, energy weighted histogram and analytic acoustic intensity method. Extending to array signal processing, the output and natural directivity of the uniform line array are introduced. Meanwhile, the array gain and vector advantage were emphasized.The noise spectrum radiated from air targets consists of line and continuous spectrum, in which the line spectrum noises are principal noise sources.In order to extract line and continuous spectrum, a new algorithm based on EMD was proposed. Some sound energy is transmitted into water when sound goes by the air-water interface, so its underwater field also holds obvious line spectrum characteristic. In low-frequency stage, there is phase different between p and v, so the sound energy leaks out between active and reactive component of acoustic intensity flux, which made conventional azimuth estimation theories biased. Based on lake and sea experiment, this dissertation probes into azimuth estimation methods of air targets using the information of underwater field, and a scheme including detection and identification of air targets has been proposed in which considered the waveband of line spectrum, the ratio of Doppler frequency offset to original frequency and the azimuth (movement changing tendency) of line frequency as the characteristic of air targets.Meanwhile the approximate azimuth information is advantaged for underwater platform to dodge and attack air targets effectually, which is helpful to improve submarines inferior position.The transient signals such as splashes sound of target’s water entry and underwater target-starting sound are very important for targets’detection. The character of targets’water entry splash sound and its waveform are studied in this dissertation which can provide a theoretical basis for the identification of water-entry sound and the forecast of impact sound’s source level.Combining the character of EMD and the advantage of acoustic intensity averager in single target azimuth estimation, mode acoustic intensity averager (MAIA) method is proposed in this dissertation. By MAIA algorithm, water-entry sound and underwater target-starting sound could be detected and their azimuth could be estimated in the strong interferences flat-form, which was helpful for transient signals detection.The vector array MVDR beam forming algorithm can estimate the omni-directional spatial spectrum effectively, but it’s difficult to detect weak targets. Combining the IMF’s narrow-band character and its advantage of weak signal detection, the vector array mode domain MVDR algorithm is proposed in this dissertation, in which the decomposed IMFs of the same order from pressure and particle velocity are used to estimate their spatial spectrum.Meanwhile applying the central frequency conception in each IMF, the central frequencies of different order are obtained by the IMF of pressure and particle velocity, and the IMF’s central frequencies are considered as the central frequency of MMVDR algorithm. The sea trial results show MMVDR algorithm can enhance the weak signal’s spatial spectrum energy effectively meanwhile keep the interference spectrum energy unchanged, and can detect the weak target effectively in strong interference, which can increase the detection distance of weak targets. The steered minimum variance (STMV) algorithm is a kind of high resolution beam-forming algorithm which holds the lower side lobe and narrower main lobe advantages.It could be concluded that its calculation cost has direct ratio with the cube of sensors number M. When the towed line array’s sensors number M is larger, the calculation cost of STMV algorithm is much more which limited its application. Considering this limitation, combining the integral function of one-order recursive filter with iterative formula of the inverse matrix, STMV iterative algorithm is proposed in the dissertation, and its calculation cost has direct ratio with M2 approximately. Simulation results showed iterative algorithm could reduce the calculation cost, meanwhile could preserve the high resolution and weak signal detection characteristics of STMV synchronously. When applied in experiment of 48 sensors line array, the calculation cost is reduced to 1/24 of original algorithm, which is much easier for its real-time realization. Meanwhile the proposed algorithm can estimate each target’s azimuth even when the sources’power differs in large scales or the sources azimuths are adjacent.更多还原