被动多目标分辨与方位估计技术研究

发布时间：2018-05-05 21:13

本文选题：阵列信号处理 + DOA估计　；参考：《西北工业大学》2016年博士论文

【摘要】：被动多目标分辨与方位估计技术研究在声纳、雷达、海洋探测等领域中有着非常重要的地位。论文系统研究了被动多目标分辨与方位估计的关键技术,主要研究内容包括:1.建立了DOA估计的阵列模型,对经典的窄带高分辨子空间算法进行了分析,研究了常用的解相干方法及快速子空间算法,并对其进行了仿真验证。针对窄带准平稳信号下KR-MUSIC算法运算量大且需要信源先验信息的问题,提出了无需信源数目的低复杂度算法,该方法利用信号的分段平稳性、结合Khatri-Rao积虚拟增加了阵列孔径。通过正交补算子可降低噪声的影响,利用循环移位矩阵降低了阵列方向矢量的维数,提高了算法的运行速度,结合全加权MUSIC理论实现了无需信源先验信息的准平稳信号DOA估计。实验仿真展示了该方法能估计的信源数超过了阵元数目,且具有较高的估计精度。2.研究了宽带非相干信号子空间法(ISSM:Incoherent Signal Subspace Method)及相干信号子空间法(CSSM:Coherent Signal Subspace Method),分析了它们的测向原理,对典型的聚焦类算法RSS及TCT进行了研究,通过仿真验证了算法的有效性。分析了宽带投影子空间正交性测试算法(TOPS:Test of Orthogonality of Projected Subspace)和频域子空间正交性测试算法(TOFS:Test of Orthogonality of Frequency Subspace)。基于对称谱理论,提出了一种新的宽带相干信号快速DOA估计算法,该方法利用Toeplitz矩阵重构,可实现不损失阵列孔径的信号解相干,结合实值转换将复矩阵映射为实矩阵,降低了子空间分解的运算量,利用噪声子空间及其共轭重新构造空间谱,根据谱对称性,在半谱内搜索即可得到信号的方位角。仿真结果表明该算法能处理宽带相干源,运算量小且具有较小的估计误差。3.研究了NC-MUSIC、实值NC-MUSIC、EPM等非圆信号DOA估计算法并进行了仿真验证。分析了非圆信号和圆信号共存下的MUSIC算法、ESPRIT算法及酉ESPRIT算法,信号共存下,信号子空间维数得到了扩展,可以分辨更多的信号,在信源过载下算法有效。针对非圆信号和圆信号共存下MUSIC算法运算量较大的问题,提出了一种快速算法,该方法将实值转换和多级维纳滤波技术应用到信号共存下,降低了该情形下MUSIC算法的运算量,利用谱峰中圆信号角度会重复出现而非圆信号角度只会出现一次的规律,可以准确区分圆与非圆信号。仿真结果验证了算法的有效性。4.研究了窄带不相关和相干信号混合下的DOA估计算法,对空间差分矩阵平方算法和空间差分矩阵特征分解算法进行了理论分析和实验仿真。为提高低信噪比下算法分辨的成功概率,提出了空间差分Mini-Norm算法,所提算法在不相关和相干信号的DOA估计中采用的都是Mini-Norm方法,算法分辨率高、对参数变化不敏感,仿真验证了新算法具有良好的性能。针对空间差分矩阵平方算法运算量大且解相干会带来阵元损失的问题,给出了一种改进算法,该方法利用新空间差分技术,解相干的阵元损失较传统的空间平滑要小,结合实值转换及线性传播算子理论,可快速获得相干信号的噪声子空间。计算机仿真实验表明所提算法估计精度较高且具有较低的计算复杂度。5.利用实值转换及空间差分技术,结合TOFS算法理论,发展了宽带不相关信号和相干信号混合下的DOA估计算法,该算法利用宽带信号各个频点协方差矩阵的厄米特性,通过酉转换可实现该频点复矩阵的实值化,通过实值化的TOFS法能迅速实现不相关信号的DOA估计。利用空间差分及Toeplitz化技术,通过实值TOFS算法,在不降低阵列孔径的条件下,可快速获得相干信号的DOA估计。仿真结果表明新算法能同时估计宽带不相关和相干信源,在信源数目大于阵元数目的情形下DOA估计有效,且具有较低的运算量。
[Abstract]:The research of passive multi-target resolution and azimuth estimation technology plays a very important role in sonar, radar, ocean detection and other fields. This paper systematically studies the key technologies of passive multi-target resolution and azimuth estimation. The main research contents include: 1. the array model of DOA estimation is established, and the classical narrowband high resolution subspace algorithm is carried out. In this paper, the common method of decoherence and fast subspace algorithm are studied, and the simulation verification is carried out. In view of the problem that the KR-MUSIC algorithm is large and needs the prior information of the source under the narrow band quasi stationary signal, a low complexity algorithm without the number of sources is proposed. This method uses the piecewise stability of the signal and combines with the Khatri-Rao. Through the orthogonal complement operator, the effect of the noise is reduced. The dimension of the array direction vector is reduced by the cyclic shift matrix, and the speed of the algorithm is improved. The quasi stationary signal DOA estimation without the prior information of the source is realized with the full weighted MUSIC theory. The experimental simulation shows that the method can be estimated. The number of sources exceeds the number of elements, and has a high estimation precision..2. studies the ISSM:Incoherent Signal Subspace Method (Method) and the coherent signal subspace (CSSM:Coherent Signal Subspace Method). The principle of their direction finding is analyzed, and the typical focusing algorithm RSS and TCT are studied. The effectiveness of the algorithm is verified by over simulation. The TOPS:Test of Orthogonality of Projected Subspace and the frequency domain subspace orthogonality testing algorithm (TOFS:Test of Orthogonality of Frequency Subspace) are analyzed. Based on the symmetric spectrum theory, a new broadband coherent signal is proposed. The algorithm uses the Toeplitz matrix reconstruction to realize the signal decoherence without loss of the array aperture. The complex matrix is mapped into a real matrix by the real value conversion. The computation of the subspace decomposition is reduced. The spatial spectrum is reconstructed by the noise subspace and its conjugate. According to the spectral symmetry, the signal can be obtained in the semi spectrum. The simulation results show that the algorithm can deal with wideband coherent sources with small computation and smaller estimation error.3.. The NC-MUSIC, real NC-MUSIC, EPM and other non circular signal DOA estimation algorithms are simulated and verified. The MUSIC algorithm, ESPRIT algorithm and unitary ESPRIT algorithm, and the coexistence of signal, are analyzed in the presence of non circular signals and circular signals. At the same time, the dimension of the signal subspace is extended to distinguish more signals, and the algorithm is effective under the source overload. A fast algorithm is proposed for the problem of the large amount of operation of MUSIC algorithm under the coexistence of non circular signals and circular signals. This method applies the real value conversion and multilevel Wiener filtering technique to the coexistence of signals and reduces the situation. The calculation of the next MUSIC algorithm, using the angle of the circular signal in the spectral peak and the angle of the circular signal, only one time, can distinguish between the circle and the non circular signal. The simulation results verify the effectiveness of the algorithm.4. and study the DOA estimation algorithm of the mixture of narrowband unrelated and coherent signals, and the square calculation of the spatial difference matrix. In order to improve the success probability of algorithm resolution under low signal to noise ratio, a spatial difference Mini-Norm algorithm is proposed. The proposed algorithm used in the DOA estimation of unrelated and coherent signals is all Mini-Norm square method. The algorithm has high resolution and is insensitive to the change of parameters. The simulation proves that the new algorithm has good performance. An improved algorithm is given for the problem that the space difference matrix square algorithm is large and coherent will cause the loss of the array element. This method uses the new space difference technique, and the loss of the decoherence array element is smaller than the traditional space smoothing, and combines the real value conversion and the linear propagator theory. On the other hand, the noise subspace of coherent signals can be quickly obtained. Computer simulation experiments show that the proposed algorithm has high estimation accuracy and low computational complexity.5. using real value conversion and spatial difference technology. Combined with the theory of TOFS algorithm, the DOA estimation algorithm with wide band uncorrelated signal and coherent signal is developed. The real value of the complex matrix of the frequency point can be realized by the unitary transformation, and the DOA estimation of the uncorrelated signals can be realized quickly through the unitary transformation. By using the spatial difference and the Toeplitz technology, the coherent signal can be quickly obtained without reducing the aperture of the array with the real value TOFS algorithm. The simulation results show that the new algorithm can simultaneously estimate the wideband uncorrelated and coherent sources, and the DOA estimation is effective in the case of the number of sources greater than the number of elements, and has a lower computation.

【学位授予单位】：西北工业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN911.7

【相似文献】