【作者】 张锴；

【导师】 梁国龙；

【作者基本信息】 哈尔滨工程大学 ， 信号与信息处理， 2012， 博士

【摘要】 声矢量传感器可以同时拾取声场中共点的声压与振速信息，它的出现为解决水下目标的探测、定位及识别等诸多问题提供了一种新的方法和手段，受到研究学者的日益关注。与标量阵算法相比，单声矢量传感器就具有目标测向能力，而声矢量阵算法的信源检测和参数估计性能也更高。随着MUSIC、ESPRIT等子空间类DOA估计方法相继被移植到矢量信号处理领域中，基于声压、振速信息联合处理的声矢量传感器空间谱估计技术已经得到学者们的广泛认可。在此背景下，本文结合常规阵列信号处理技术，进一步利用声矢量传感器输出的声场信息，对单/双声矢量传感器的高分辨方位估计、声矢量阵的信源数目估计、声矢量阵的高分辨方位估计及阵型校正等几方面内容进行深入、系统的理论和实验研究。基于单声矢量传感器的测向问题一直是学者们的研究热点，本文首先讨论了基于单矢量传感器的Bartlett、Capon和MUSIC等几种算法的估计性能，在建立数学模型的基础上重点研究了单矢量传感器MUSIC算法的方位估计统计特性及性能边界问题，详细分析了通道幅相特性对MUSIC算法性能的影响，并针对幅度误差的影响提出了一种基于通道功率归一化的校正算法，理论分析、仿真实验以及湖试数据处理结果均验证了本文算法的有效性。同时针对小尺度声矢量阵的相干源分辨问题，提出一种基于空间平滑思想的二元声矢量阵MUSIC算法，该算法依靠两个声矢量传感器可实现对相干信号源的分辨。信源数目的精确估计是保证高分辨方位估计算法实现的前提条件，本文在讨论AIC、MDL、GDE和CCT等几种经典算法性能的基础上，通过将声压、振速或解析振速信息进行联合处理，结合正则相关技术提出了三种基于声矢量阵的信源数目估计算法—VCCT算法。仿真结果表明：与CCT算法相比，VCCT算法显著提高了可估计的信源数目上限，而且在信源空间位置变化的情况下，VCCT算法的信源数目估计性能具有稳健性。与标量阵算法相比，声矢量阵高分辨方位估计算法所需运算量较大，基阵的使用要求也更高。针对上述问题，本文结合空间拟合思想提出了一种新的高分辨方位估计算法—VSSF算法。该算法在估计性能上接近于声矢量阵MUSIC算法并优于ESPRIT算法，但运算量约为MUSIC算法的一半。为了提高算法的分辨能力，本文将VSSF算法推广为波束域处理的BVSSF算法，分析表明：该算法比VSSF算法的信源估计数目有所减少，但具有更高的分辨力。针对宽带信号情况，本文同时给出了两种宽带高分辨方位估计算法—WIVSSF算法和WFVSSF算法。分析表明：WIVSSF算法具有非相干源能力，WFVSSF算法具有非相干源与相干源分辨能力。针对阵元位置存在误差时VSSF算法性能恶化问题，本文提出一种新的声矢量阵阵型校正算法，该算法利用单矢量传感器MUSIC算法估计disjoint辅助源的方位，并以此构造完备方程组求解误差矩阵。仿真结果表明：一定条件下本文算法仅依靠2个disjoint辅助源就可以实现阵元位置误差的有效估计，与在同等条件下文献算法相比其估计精度要高于后者一个数量级。更多还原

【Abstract】 The Acoustic Vector Sensor (AVS) can measure both pressure and particle velocity ofacoustic field at a point in space，which provides a new technique for underwater targetdetection, location and identification. In recent year, AVS signal processing has been drawnmuch attention. Compared with conventional Pressure Sensor Array (PSA) signal processingalgorithm, the Single Acoustic Vector Sensor (SAVS) can achieve Direction-of-Arrive (DOA)estimation for object, the better performance can be obtained from aspects of sourcesdetecting and parameter estimating based on the Acoustic Vector Sensor Array (AVSA) signalprocessing algorithm also be better. With the MUSIC, ESPRIT algorithms explanted to theAVSA signal processing, the spatial spectrum estimation technique based on the pressure-velocity combined processing using AVSA has been accepted widely. Combining theconventional array processing technique, the thesis makes use of the information exportingfrom AVS to research mainly three parts. They focus on the problems such as high resolutionDOA estimation using the single/duality AVS, sources number estimation, high resolutionDOA estimation and array calibration using the AVSA.The DOA estimation using SAVS is always an important topic in acoustic signalprocessing field. The paper first debates the performance of Bartlett, Capon and MUSICalgorithms based on the SAVS. In allusion to MUSIC algorithm, the paper investigates thestatistical character and performance boundary of DOA estimation particularly, analyses theinfluence on algorithm caused by intensity and phase error, meanwhile a novel approach usingchannels power unitary is proposed to eliminate the influence leading by intensity error. Thevalidity of the approach is validated through theoretic analysis, numeric simulation and lakeexperiment data processing. In allusion to small-size AVSA, the thesis also propose a MUSICalgorithm combined the principle of spatial smoothing for coherent sources resolution, whichis only depended on duality AVSA.The foreknowing of source number is the precondition for high resolution DOAestimation. After the paper introduces some classical source number estimation approach suchas AIC, MDL, GDE and CCT algorithm, three novel approaches for source number estimationare proposed (VCCT) combining pressure-velocity combined processing with CCT algorithm.The result of simulation indicates that, comparing with CCT algorithm using PSA, VCCTalgorithm using VASA evidently increases the upper limit of source number estimation.Further more, the performance of VCCT algorithm also has robustness when the source at different location in space.The high resolution DOA estimation using AVSA has more quantity in operation andmore complicated in application than PSA. In allusion to this problem, the paper proposed anew high resolution DOA estimation algorithm using AVSA based on the principle ofsubspace fitting (VSSF). The performance of proposed measure is closer to the MUSICalgorithm and better than ESPRIT algorithm. But the computation of operation is about halfof MUSIC algorithm’s. The VSSF algorithm is also explanted to the beam field and widebandfield. Meanwhile the paper produces the BVSSF algorithm, WIVSSF algorithm and WFVSSFalgorithm. The results of simulation indicates that, the BVSSF algorithm has the higherresolution in despite of reducing the upper limit of source number. The WIVSSF algorithmcan distinguish the incoherent sources and WIVSSF algorithm can distinguish the coherentsources.A novel approach for array calibration using the AVSA is proposed because theperformance of VSSF algorithm is depraved when array location exist error. The proposedapproach makes use of DOA estimation of disjoint source by MUSIC algorithm based onSAVS, and constitutes the equation groups to calculate the error matrix. Simulation resultindicates that, the approach, only depending on two disjoint sources, is proved to be valid forarray calibration, and its precision is more one order of magnitude than the literatureapproach.更多还原