基于粒子滤波算法的声矢量传感器DOA跟踪估计研究

发布时间：2018-10-26 08:13

【摘要】：声矢量传感器(Acoustic Vector Sensor,AVS)技术可应用在通信定位、声呐、故障源定位、雷达以及生物医学等众多的国民经济以及军事国防领域，同时声矢量传感器能够同步共点地测量某点处的声压和质点振速的矢量信息，获取多维声信号信息，进而能够用于分析处理的信息更多，因此，声矢量传感器技术以及声矢量传感器信号处理技术一直是备受关注的研究方向。近些年来，基于声矢量传感器的阵列信号处理技术的研究主要集中在空间谱估计上，而空间谱估计算法都是建立在假设目标在观测时间内是静止的这一前提之下，在实际环境中，信号源通常是运动的，致使目标的波达方向(Direction of Arrival，DOA)在观测时间内静止的假设不再成立，针对动态目标的情况，目标入射方位估计算法会造成较大的估计误差。随着人们对目标测向的精确度和实时性要求越来越高，静态波达方向估计算法不能满足日益增高的要求，因此研究用于动态目标的DOA估计跟踪算法具有重要意义。本文主要研究了基于粒子滤波算法以及其改进算法的声矢量传感器DOA跟踪估计算法。首先对声矢量传感器的阵列流型和测量模型的建立进行推导，对粒子滤波算法和贝叶斯估计理论进行详细的介绍。其次，研究了基于粒子滤波算法的DOA估计跟踪算法，并且在该算法的基础上，针对俯仰角和方位角的相对独立性，提出了分别评价样本权重的改进粒子滤波算法，并将其应用于方位跟踪方面。再次，利用空间谱估计理论中的经典高分辨MUSIC(Multiple Signal Classfication, MUSIC)算法的空间谱与似然函数的相似性，提出了改进似然函数的粒子滤波跟踪算法。然后，基于四元数理论结合导向矢量的长矢量模型的推导，建立声矢量传感器导向矢量的四元数模型，给出了四元数MUSIC算法的谱估计式，利用其作为粒子权重的评价函数，提出了一种改进的DOA跟踪算法以及改进粒子滤波算法的跟踪算法。最后，，通过MATLAB仿真实验验证了本文提出的跟踪算法可实现目标方位的跟踪估计，并且具有较好的跟踪性能。本文主要针对声矢量传感器研究了DOA跟踪问题，期待本文的研究有助于声矢量阵列信号理论的发展和实际工程的应用。在今后的工作中，在声矢量传感器DOA跟踪领域，相关研究也需继续进行。
[Abstract]:Acoustic vector sensor (Acoustic Vector Sensor,AVS) technology can be used in many fields of national economy and military defense, such as communication location, sonar, fault source location, radar, biomedicine and so on. At the same time, the acoustic vector sensor can synchronously measure the vector information of sound pressure and particle vibration velocity at a certain point, and obtain multi-dimensional acoustic signal information, which can be used to analyze and process more information. Acoustic vector sensor technology and acoustic vector sensor signal processing technology have been paid close attention to. In recent years, the research of array signal processing technology based on acoustic vector sensor is mainly focused on spatial spectrum estimation, and spatial spectrum estimation algorithms are based on the assumption that the target is stationary in observation time. In the actual environment, the signal source is usually moving, so that the assumption that the target's direction-of-arrival (Direction of Arrival,DOA) is stationary during the observation time is no longer true, for the case of the dynamic target, The estimation algorithm of the incident azimuth of the target will cause a large estimation error. With the increasing demand for accuracy and real-time of target direction finding, the static DOA estimation algorithm can not meet the increasing requirements. Therefore, it is of great significance to study the DOA estimation and tracking algorithm for dynamic targets. This paper mainly studies the DOA tracking and estimation algorithm of acoustic vector sensor based on particle filter and its improved algorithm. Firstly, the array flow pattern and measurement model of acoustic vector sensor are deduced, and the particle filter algorithm and Bayesian estimation theory are introduced in detail. Secondly, the DOA estimation and tracking algorithm based on particle filter algorithm is studied, and based on this algorithm, an improved particle filter algorithm is proposed to evaluate the weight of samples, aiming at the relative independence of pitch angle and azimuth angle. It is applied to azimuth tracking. Thirdly, using the similarity between spatial spectrum and likelihood function of classical high-resolution MUSIC (Multiple Signal Classfication, MUSIC) algorithm in spatial spectrum estimation theory, a particle filter tracking algorithm with improved likelihood function is proposed. Then, based on the quaternion theory and the derivation of the long vector model of the guidance vector, the quaternion model of the guidance vector of the acoustic vector sensor is established, and the spectrum estimation formula of the quaternion MUSIC algorithm is given, which is used as the evaluation function of the particle weight. An improved DOA tracking algorithm and an improved particle filter algorithm are proposed. Finally, the MATLAB simulation results show that the proposed tracking algorithm can achieve the target azimuth estimation and has good tracking performance. In this paper, the DOA tracking problem for acoustic vector sensors is studied, and it is expected that the research in this paper will be helpful to the development of acoustic vector array signal theory and the application of practical engineering. In the future work, in the field of acoustic vector sensor DOA tracking, the related research also needs to be continued.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN911.7;TN713

【参考文献】