基于稀疏贝叶斯的相关信号DOA估计

发布时间：2018-05-25 04:07

本文选题：稀疏贝叶斯学习 + 稀疏重构　；参考：《电子科技大学》2017年硕士论文

【摘要】：目前阵列测向系统面临低信噪比、相关信号、多径效应等复杂环境,传统测向方法,如子空间类方法,已经难以实现准确测向。现有基于信号空域稀疏性的稀疏重构方法表现出一些显著的优势,但是这些稀疏重构方法受到一些因素的限制,存在一定的缺陷。本文基于信号空域稀疏性,通过稀疏贝叶斯框架研究点源中相关信号和分布源中相干分布源的快速高精度DOA估计问题。主要研究内容和创新如下:1、研究相关信号的点源稀疏表示模型及传统的相干分布源模型。分析基于0?范数的稀疏重构和最大后验概率估计的内在联系;着重探讨了贝叶斯方法中的稀疏贝叶斯学习方法,给出其理论框架,并分析了贝叶斯方法在阵列测向方面的相关性质。2、基于稀疏贝叶斯框架,提出基于阵列二阶项的DOA估计方法。通过协方差差分技术抑制噪声影响,使用一种新的空域采样方式形成过完备字典,并将其转化为稀疏模型。在稀疏贝叶斯学习的基础上,通过有效的DOA搜索方法估计DOA。给出了所提方法的计算复杂度分析,并与L1-SVD等方法进行了性能对比分析,验证了所提方法的正确性和有效性。3、在波束空间,提出一种基于稀疏贝叶斯学习的快速高精度DOA估计方法。将阵列输出映射到波束空间,通过较大角度间隔的过完备字典稀疏表示协方差。联合稀疏贝叶斯学习和DOA精细化估计方法进行点源相关信号的高精度DOA估计。给出了协方差稀疏表示通过增加快拍数提高信噪比的可行性分析及所提方法的计算复杂度分析,并与L1-SRACV等方法进行了性能对比分析,验证了所提方法的正确性和有效性。4、提出一种基于稀疏贝叶斯学习的一维相干分布源参数估计方法。建立相干一维分布源的稀疏表示模型,通过对空域和角度扩展域进行采样形成广义阵列流行矩阵。在此基础上,提出基于稀疏贝叶斯学习的相干分布源参数估计方法,通过粗估计和精估计获得分布源的角度参数估计。给出了所提方法的计算复杂度分析,并与DSPE等方法进行了性能对比分析,验证了所提方法的正确性和有效性。
[Abstract]:At present, the array direction-finding system is faced with complex environments such as low signal-to-noise ratio (SNR), correlation signal, multipath effect and so on. Traditional direction-finding methods, such as subspace methods, have been difficult to achieve accurate direction finding. The existing sparse reconstruction methods based on signal spatial sparsity show some obvious advantages, but these sparse reconstruction methods are limited by some factors and have some defects. In this paper, based on the spatial sparsity of signals, the problem of fast and high accuracy DOA estimation for correlated signals in point sources and coherent distributed sources in distributed sources is studied by using sparse Bayesian framework. The main contents and innovations are as follows: 1. The point source sparse representation model and the traditional coherent distributed source model are studied. Analysis based on 0? This paper mainly discusses the sparse Bayesian learning method in Bayesian method, and gives the theoretical framework of the sparse Bayesian learning method, which is related to the sparse reconstruction of norm and the estimation of maximum a posteriori probability. The related properties of Bayesian method in array direction finding are analyzed. Based on sparse Bayesian framework, a second order DOA estimation method based on array is proposed. Using covariance difference technique to suppress noise effect, a new spatial sampling method is used to form an overcomplete dictionary, which is transformed into a sparse model. Based on sparse Bayesian learning, DOA is estimated by an effective DOA search method. The computational complexity analysis of the proposed method is given, and the performance of the proposed method is compared with that of L1-SVD, which verifies the correctness and validity of the proposed method in beamspace. A fast and high precision DOA estimation method based on sparse Bayesian learning is proposed. The array output is mapped to beamspace and covariance is represented sparsely by overcomplete dictionaries with large angular intervals. Combining sparse Bayesian learning and DOA refinement estimation, a high accuracy DOA estimation of point source correlation signals is presented. The feasibility analysis of covariance sparse representation to increase SNR by increasing the number of beats and the computational complexity analysis of the proposed method are given, and the performance of the proposed method is compared with that of L1-SRACV and so on. The validity and validity of the proposed method are verified. 4. A method for estimating the parameters of one-dimensional coherent distributed sources based on sparse Bayesian learning is proposed. The sparse representation model of coherent one-dimensional distribution source is established and the generalized array prevalence matrix is formed by sampling the spatial domain and angular spread domain. On this basis, a sparse Bayesian learning method for estimating the parameters of coherent distributed sources is proposed. The angle parameters of distributed sources are estimated by rough estimation and refined estimation. The computational complexity analysis of the proposed method is given, and the performance of the proposed method is compared with that of DSPE and so on. The correctness and effectiveness of the proposed method are verified.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN911.23

【相似文献】