【作者】 张东兴；

【导师】 索莹；

【作者基本信息】 哈尔滨工业大学 ， 信息与通信工程， 2020， 硕士

【摘要】 作为阵列信号处理(Array Signal Processing)的一个核心内容,波达方向DOA(Direction of Arrival)估计技术广泛应用于电子侦察与对抗、生物医学成像、声纳和通信系统等领域。实际的阵列测向系统中由于各种不可控因素广泛存在多种误差,例如阵列幅相不一致、阵元互耦效应和阵元位置不准确等,导致阵列流形出现一定程度的偏差,致使现有的大多数DOA估计算法性能产生显著的负面影响。本文基于空间谱估计技术和压缩感知(Compressed Sensing,CS)相关理论,利用DOA在空域的稀疏性和稀疏贝叶斯学习SBL(Sparse Bayesian Learning)方法实现波达方向和阵列误差参数的联合估计。在传统的基于压缩感知(的系统模型中,空间角度域划分为均匀的离散网格,离散网格(offgrid)的量化误差限制了估计性能的进一步提高。本论文主要针对常见的三种阵列模型误差条件下的DOA估计问题进行了研究,主要研究内容概括如下:首先,研究,分析基于稀疏贝叶斯学习理论的DiOA估计,表明该方法相对于传统的DOA估计的优越性。其次,从理论分析得出基于稀疏,贝叶斯学习,方法的DOAs估计方法存在两个不足:噪声方差估计有偏和空域离散超完备模型所引入的量化误差。针对噪声方差估计不准确利用信息论方法和相关论文辅助得到适用于稀疏贝叶斯学习方法的无偏噪声方差估计;针对空域离散超完备模型所引入的量化误差问题利用去网格化格稀疏模型,联合估计off-grid参数和采样网格上DOAs实现高精度的波达方向估计。然后,对于测向系统中的阵元间互耦效应,基于上述的改进贝叶斯学习方法,建立新的观测模型,提出了一种基于阵列互耦误差的稀疏去网格DOA估计算法,迭代更新了包括DOAs、阵列误差向量、噪声功率参数和off-grid向量等在内的未知参数。将基于稀疏去网格化贝叶斯学习方法理论又进一步拓展,实现波达方向和阵列幅相误差联合估计以及波达方向和阵元位置误差联合估计。最后,仿真结果表明,与现有方法相比,在存在阵列误差条件下,本论文所提出的算法显著提高了DOA估计性能。更多还原

【Abstract】 In the field of array signal processing,DOA(Direction of Arrival)estimation is an important research hotspot,which is widely used in radar,communication system,sonar and biomedical imaging.In the actual array system,there are a wide range of error errors,such as mutual coupling effect of array elements,array amplitude and phase inconsistency and array element position inaccuracy,which will lead to a certain degree of deviation of array flow pattern,resulting in significant negative impact on the performance of most existing DOA estimation algorithms.Based on the relevant theories of compressed sensing and spatial spectrum estimation,this paper USES the sparse and sparse Bayesian learning methods of DOA in the spatial domain to realize the joint estimation of wave arrival direction and array error parameters.In the traditional system model based on compressed sensing(CS),the spatial domain is discretized into grids,and the quantization error of discrete grids limits the further improvement of estimation performance.In this paper,DOA estimation under three common array model errors is studied.The main research contents are summarized as follows:Firstly,DOA estimation based on sparse Bayesian learning method is studied and analyzed,indicating its superiority over traditional DOA estimation.Secondly,it is concluded from theoretical analysis that the DOA estimation method based on sparse Bayesian learning method has two shortcomings: bias of noise variance estimation and quantization error introduced by spatial discrete supercomplete model.Aiming at the inaccuracy of noise variance estimation,the unbiased noise variance estimation suitable for sparse Bayesian learning is obtained with the help of information theory and related papers.Aiming at the quantization error problem introduced by spatial discrete supercomplete model,the de-meshing lattice sparse model is used to jointly estimate off-Grid parameters and DOAs on the sampling grid to realize high-precision estimation of the arrival direction.Then,for the mutual coupling effect between array direction finding system of yuan,improved Bayesian learning method based on the above,a new observation model,this paper puts forward a kind of sparse array mutual coupling error off-grid DOA estimation algorithm,iterative updated including DOAs,array error vector and the parameters of the noise power and off-grid vector of unknown parameters.Based on sparse de-meshing Bayesian learning method,the joint estimation of wave arrival direction and array amplitude and phase error and the joint estimation of wave arrival direction and array position error are realized.Finally,the simulation results show that compared with the existing methods,the proposed algorithm significantly improves the DOA estimation performance in the presence of array error.更多还原