【作者】 陈阳；

【导师】 惠俊英；

【作者基本信息】 哈尔滨工程大学 ， 水声工程， 2010， 博士

【摘要】 阵列信号处理在很多应用领域具有重要作用。随着传感器和阵列技术的发展，与之相应的阵列信号处理成为近几十年的热点，受到广泛的关注。本文从阵列信号处理的目标检测、方位分辨和估计的角度，对现有算法做出改进，以提高算法性能。空域匹配滤波是最大信噪比准则下的最优处理器，也是单目标白噪声情况下似然估计的等价处理器。利用最小均方误差准则下的维纳滤波对阵列快拍进行滤波，可以一定程度上抑制噪声。为了在空域进行噪声抵消，利用二阶锥规划实现维纳滤波。对算法进行仿真分析，并在单目标白噪声条件下与最大似然估计(Maximum Likelihood Estlmatlng，MLE)和Bar-tlett波束形成进行比较。空域维纳滤波器的方位估计的信噪比门限要低于MLE，其低信噪比下的检测性能优于Bartlett波束形成，因而是一种良好的波束形成器。最小方差无畸变响应(Mlnlmum、1arlance[)istortionless Response，MvDR)是波束输出能量最小意义上的最优波束形成。对MvDR的约束条件进行分析，将其转换为二阶锥的形式，通过将，J，范数最小条件改为，J。范数最小条件，提出基于，m范数约束的最小方差无畸变响应(，Jmnorm constlraIntMinimumv1ar。lancei)istortionless Response，，lm．MVDR)波束形成器。，lm．MvDR的权向量波束能更好地抑制干扰，因而，lm．MvDR的方位分辨信噪比门限低于MvDR。对比分析MvDR和，l一．MvDR的稳健性，当阵列存在失配时，MvDR和，lm．MvDR的性能受影响，随着失配的增大，，lm．MvDR比MvDR退化要快，因而，lm．MvDR的分辨能力是以损失对失配的稳健性为代价的。将宽带导向最小方差波束形成fSteered Mlnlmumv1arlance，sTMv0算法扩展到矢量阵，并与基于声矢量传感器均匀直线阵的：Bai~tlett、非相干最小方差(ncoheren!，Mlnlmum1arlance，IcMv)和空间重采样相干子空间最优(Spatially Resampled Minimumv1arlance，SRMv)波束形成算法进行比较：分析标量阵与矢量阵指向性、抗左右舷模糊、主波束宽度和旁瓣级、空间欠采样，方位分辨力、相关信号源的分辨能力以及强干扰下的弱目标检测能力。最后利用这几种矢量波束形成算法对三次海试数据进行了分析。当目标快速运动时，目标方位的改变会引起互谱密度矩阵的模糊，影响波束形成性能。因而，收敛快的sTMv波束形成与收敛慢的IcMv相比，方位谱主瓣更窄，旁瓣更低。为了充分发挥线谱信噪比高的特性，提高三维方位历程检测线谱目标的能力，提出了频率方差加权波束形成检测器。首先分析了短时傅里叶变换瞬时频率方差估计，给出了它的理论解，并通过Monte car0仿真验证了其正确性。当线谱的谱级信噪比超过一定门限时瞬时频率方差为零。用频率方差对波束能量进行加权，线谱目标方位波束因输出信号的瞬时频率方差较小得到增强，而其他方位波束被抑制，从而有效提高对线谱目标的检测能力。然后通过仿真频率方差加权常规波束形成(varlance of lnst3mtaneous frequencv．conventional Beaanib~"ruing Beamibrmlng，vIF．cBF)和频率方差加权导向最小方差波束形成(v1arlance of instantaneous frequency．Steered Minimumv1arlance BeamIbianlng，vIF．sTMv)，分析了频率方差加权波束形成抑制无线谱强干扰，增强线谱目标的机理。还比较了倒数加权与指数加权的特性。最后海试数据处理验证了本文提出的vIF．sTMv检测器，处理结果表明，vIF．sTMv检测器可以有效的提高线谱目标的检测能力。更多还原

【Abstract】 Array signal processing plays a very important role in various applicationfields With the development of sensor and array technology,it has been a populartopic in recent years In this paper,several algorithms are improved to enhancetheir capabilities for target detecting，bearing resolution and DOA estimatingSpatial match filter is the optimal processor under maximum signal to noiseratio(SNR)criterion in mono-source and white noise condition The noise can besuppressed to a certain extend by tilting the array snapshots with Wiener filterwhich is under minimum mean-square error criterion To carry out spatial noisecmaceling，the Wiener filter is constructed through Second-Order ConeProgramming(SOCP)The algorithm is then compared with maximum likelihoodestimating(MLE)and Bartlett beamforming through simulation The SNRthreshold of Spatial filter is lower than maximum likelihood method in DOAestimation，and the detection performance is better at low SNR So it is a excellentprocessor in mono-source mad white noise conditionMinimum Variance Distortionless Response(MVDR)is the optimalprocessor under minimum power of beam output and distortionless responsecriterion Its constraints can be transformed to second-order cone forms L,norlTl constraint Minimum Variance Distortionless Response。-MVDR)isproposed by minimizing L,norlTl instead of L2 norlTl of the power minimizingconstraint in MVDR The weights generated by L,-MVDR repress interferesbetter than MVDR，which is the cause of a better bearing resolution Analyse therobustness of both algorithms comparatively The disturbance of array modelaffects beamformings’behaviors With the increase of the disturbance，L,-MVDR deteriorates more rapidly than MVDR So L,-MVDR sacrifices itsrobustness to gain better resolution ability．Acoustic vector-sensor array STeered Minimum Variancebeamforming(VSTMV)is propounded A comparative analysis of Bartlett Incoherent Minimum Variance(ICMV)mad Spatially Resmnpled MinimumVariance(SRMV)beamforming algorithms based on vector uniform line aiTa iscarried out Then several aspects are analysis such as directivity,port／staxboarddistinguishing，width of mainlobe，sidelobe levee spatial undersamplingperformance，bearing resolution，resolution of correlated sources mad weak sourcedetecting ability under s~ong interferes Finally sea trial real data is maalysed withthese algorithms comparatively Cross-spe~rM density matrix estimated will notbe accurate with the direction of the target changing caused by its moving，whichleads to beamforming performance’S degenerating So that，STMV with fasterconvergent speed has better performanceVariance of Instantaneous Frequency-Beamforming detector(VIF-BF)isintroduced to improve the detecting ability of targets that radiate line spectrumsignal for three--dimensional time--bearing display,by taking advantage of highSNR prope~y of line spectrum First，the form for the variance of instantaneousfrequency estimating through short-time Fourier Transform(STFT)is derived．and is verified with Monte Carlo simulation The beamforming outputs of thebearings where the targets that radiate line spectrum signal are will bes~enghened，and vice versa Therefore the detecting capacity is enhancedVariance of instantaneous~equency-CBF beamforming detector and variance ofinstantaneous~equency-STMV beamforming detector are simulation as aexample The behavior of multiplicative inverse method and exponentiN weightmethod is also studied Its validity is proved by three sea trial更多还原