【作者】 程彬彬；

【导师】 杨士莪；

【作者基本信息】 西北工业大学 ， 水声工程， 2007， 博士

【摘要】 众所周知，声波兼有标量场和矢量场。与传统的声压水听器相比，矢量水听器既能获得声场中的声压这一标量信息，又能获得声场中质点振速这一矢量信息，更多的信息量必然带来矢量水听器更好的信号处理效果。单个的声压水听器由于只能获得声场中的声压这一标量信息，因此不具备指向性，所以不能进行信号方位的估计，而单个矢量水听器由于能够以空间共点、时间同步测量声场中的声压和振速信息，因此具备信号方位估计的能力。本文以二维压差式矢量水听器为研究对象，对单个的二维压差式矢量水听器信号方位估计做了理论推导、计算机仿真和试验数据分析。主要内容如下：1．回顾了矢量水听器及其信号处理技术的发展，对同振型矢量水听器及压差式矢量水听器工作原理进行了介绍，并建立了单个同振型及压差式矢量水听器信号模型，给出了不同组合下的二维矢量水听器波束形成仿真。2．对声场中声压和质点振速之间的相关性做了分析，提出了基于单个二维压差式矢量水听器的多目标分辨算法，该算法有效地利用了平面波声场中声压和质点振速的相关性，在声压及振速分量之间形成一组相关方程，通过对这组相关方程的优化来实现对信号方位角度的估计。文中针对窄带非相关信号、宽带信号和相干信号下的相关方程分别进行了推导。同时对已知干扰方位角度情况下的多目标分辨也给出了推导，其中目标信号包括窄带和宽带非相关信号。3．首次对各向同性白噪声条件下的二维压差式矢量水听器声压和振速之间相关方程处理增益进行了推导和分析，并对不同信号方位角度下的相关方程误差进行了仿真。推导是在窄带信号条件下获得的，但其结果在宽带信号条件下也是成立的。在信号个数较少的情况下，可以根据推导结果选择误差相对较小的相关方程来进行优化求解。4．针对本文声压和振速之间非线性相关方程组，提出运用遗传算法、差异进化算法及混合差异进化算法进行优化求解。对三种算法基本原理及步骤进行了介绍，并针对本文中要解决的多参数优化问题，对这三种方法的收敛速度、收敛精度及运算时间进行了比较。5．针对本文所运用的基于解相关方程的信号方位估计方法进行了仿真试验，并运用试验数据对算法进行了验证。仿真以两信号分辨为主，分别对窄带非相关信号、宽带非相关信号和窄带相干信号下的信号方位分辨进行了仿真，同时对多于两目标情况下的方位估计误差进行了仿真。利用西北工业大学航海学院消声水池试验数据得出了矢量水听器在不同频率下的指向性图，并对单个信号源方位估计运用声能流方估计法进行了验证；利用冯家山水库试验数据对两个非相关窄带信号、两个非相关宽带信号以及已知干扰方位条件下的信号方位估计算法进行了验证；利用舟山海试数据对实际中的宽带信号多目标分辨算法和已知干扰方位下的目标方位估计算法进行了验证。更多还原

【Abstract】 It is well known that sound wave has both scalar quantity field and vector quantity field. Compared with the pressure hydrophone, vector transducer can provide both the information about pressure (scalar quantity) and particle velocity (vector quantity) of the sound field, which offers more possibilities for signal processing. A single pressure hydrophone is omni-directional and can not be used to estimate the DOA of acoustic signal. But single vector transducer can be used to measure the pressure and particle velocity at some point of acoustic field simultaneously, so a single vector transducer can achieve the estimation of DOA. In this thesis, the theoretical deduction, computer simulation and analysis of experimental data, about estimation of multi-source parameters with single vector transducer for pressure gradient, transducer, using method of correlation equations are given. The main contents are:1. The development of vector transducer and corresponding methods of signal processing is reviewed. The working principle of co-vibrating type and pressure-gradient type of vector transducers are studied, and their mathematical models are established. The direction patterns of 2-dimensional vector transducer for different type of combination are given by simulation.2. The correlation of pressure and particle velocity of acoustic field is studied. A multi-sources distinguishing method is presented. A series of correlation equations are obtained for the correlating of pressure and particle velocity in plane wave, and DOA of sound sources are obtained by solving this system of equations. In this thesis, the system of correlation equations is solved respectively for narrowband uncorrelated signal, broadband signal and correlated signal are studied. The multi-sources distinguishing method with known direction interfere is presented, for both uncorrelated narrowband and broadband cases.3. It is the first time to study the gain in SNR using correlation of pressure and particle velocity for two-dimensional pressure-gradient vector transducer, under background of isotropic white noise, and the accuracy of solution of correlation equations at different arrival angel is simulated. Although the conclusion is deduced at narrowband case, but it keeps force also for broadband signal. According to the study, correlation equations with smaller error should be chosen to estimate the DOA in case of less sources. 4. GA (Genetic Algorithms), DE (Differential Evolution) and HDE (Hybrid Differential Evolution) are applied on the optimization of non-linear correlated function of pressure and particle velocity, the principle and steps of the three methods are introduced, the convergence speed, convergence precision and costing time of three methods are compared based on the optimization problem which needs to be solved in this thesis.5. Computer simulation and experimental data processing are applied with the DOA estimation based on solving correlation functions. Computer simulation is given mainly for the case of two sources, including uncorrelated narrowband signal, broadband signal and correlated signal. The estimation error of DOA in case of more than two sources is also provided. Directional pattern of vector transducer at different frequency is measured in anechoic tank of Northwestern Polytechnical University, and direction estimation of single source is verified by method of sound energy flux. Direction estimation method of two uncorrelated narrowband signals, broadband signal and sound signal with known direction interference are verified by experimental data of getting from experiment carried in Feng-Jiashan reservoir. Multiple broadband sources distinguishing and DOA estimation of signal interfered by known direction angle interference is verified by experimental data getting from experiment carried in sea area of Zhou-Shan.更多还原