【作者】 殷吉昊；

【导师】 陈天麒；

【作者基本信息】 电子科技大学 ， 电路与系统， 2007， 博士

【摘要】 现实环境中很多信号表现出非平稳统计性质,基于平稳窄带假设的传统阵列处理方法在分析处理这类信号时受到限制。在阵列空域处理和非平稳信号处理方法的基础上,将阵列信号处理方法与非平稳信号分析工具相结合,实现非平稳阵列信号多参量估计,是当前阵列信号处理前沿课题之一。本文以多种典型的非平稳调频、调相雷达信号为研究对象,围绕信号性质分析、非平稳信号阵列建模与时频空多参量联合估计、阵列校正等问题展开深入研究,主要研究工作及贡献有:1.在总结了相位编码信号谱相关性质的基础上,提出了一种相位编码信号循环频率和方位角联合估计新方法,与通常基于循环频率先验知识的循环平稳信号阵列处理方法相比,所提算法不但具备较强信号选择性和低信噪比到达角高分辨估计等优点,而且还可同时实现信号循环频率估计,并有效避免了通常计算信号谱相关所带来的大运算量。2.根据短时傅立叶变换下相位编码信号时频结构的分析,提出了一种相位编码信号调制参数与空域二维到达角联合估计新方法。该算法除能正确实现信号空域信息估计外,还能够精确体现不同调制样式的多分量相位编码信号时频域特征和调制特征,以较高精度实现多分量相位编码信号载频、子码宽度和编码序列参数的估计。3.提出了分数阶傅立叶变换(FRFT)——相干信号子空间方法,该方法借鉴了平稳宽带阵列信号相干信号子空间类方法中的自聚汇思路,同时结合了宽带线性调频信号在FRFT域特有的滤波特性,通过FRFT域波束空间协方差矩阵的特征分析实现了宽带线性调频信号方位角估计。该方法除了具有有效积累信号频带内的能量、提高参数估计的信噪比之外,还具有测向精度不受信号调制参数估计偏差影响以及快速收敛等优良性能。4.针对时频-相干信号子空间方法运算量大且不能直接处理相干信号的缺点,提出了一种宽带相干线性调频信号初始频率、调频斜率及二维到达角联合估计算法。该算法在估计宽带相干信源过程中,不需要在高维参数空间中设计聚汇或内插时频矩阵,也不需多维搜索,计算复杂度相对较小。5.基于正弦调频信号Jacobi-Anger级数表示,提出了一种高效的正弦调频信号调制参数及二维到达角联合估计新算法。所提出的算法不仅利用了信号级数表示后的性质来提高二维到达角估计性能,而且还辅助以信号二维到达角估计信息,实现多分量正弦调频信号间各自的基本调制参数分离和提取,进而获得了多分量正弦调频信号基本调制参数估计。算法的具体实现过程及相应的仿真结果,均展现出阵列信号处理框架下多参量联合估计的优越性能。6.研究了非平稳信号阵列误差校正问题,在到达方向依赖的阵元复增益条件下,基于宽带线调信号阵列流形的性质分析,提出了一种宽带线调信号二维角估计新方法。该方法摆脱了传统阵列校正算法中多维非线性最优化思路,具有计算量小,到达角估计可以闭式解析估计,不需额外的误差估计与补偿等处理,即可消除阵元幅相不一致误差对到达角估计的影响等一系列优良性能。更多还原

【Abstract】 In the real world, many signals whose frequency contents evolve with time are statistically nonstationary. The traditional methods in array signal processing, based on the narrowband and stationary assumptions, may fail to treat with these nonstationary signals. The theme of this dissertation is to estimate spatial and time-frequency/modulation parameters of some nonstationary radar signals, with a novel approach that combines nonstationary signal processing with array signal processing. And in the dissertation, we focus on a few aspects of nonstationary signal array processing, such as array signal modeling, signal’s property analysis, multi-parameter estimation and array error calibration. The main contributions of the dissertation are summarized as follows:1. The spectral redundancy (also called cylcostationary) property of phase-coded signals(PCSs) is first introduced, then an effective method is proposed for cylic-frequency and DOA estimation of PCSs without a prior. Compared with some existing algorithms, the proposed method can perform signal selective DOA estimation well, and get signals’ cylic-frequencies at little computational cost.2. Based on the PCSs’ time-frequency structure analyzed by STFT, a novel algorithm is presented, not only to estimate 2-dimensional angle of arrive (2-D AOA) of PCSs, but also to obtain all basic modulation parameters (i.e. carrier frequency, symbol time and code sequence) estimation for multi-component PCSs with different modulation type.3. For the problem of direction finding of wideband chirp / wideband linear frequency modulation (WLFM) signals, a new method called FRFT-CSSM is achieved, by jointly exploiting the LFM signal’s selectivity in FRFT domain and auto-focusing idea of the conventional CSSM. Except for the merit of accumulating the whole signal band’s energy to improve the singal to noise ratio, the proposed algorithm is also insensitive to initial estimates and converges fast to the true DOA value.4. To reduce computational load of the time-frequency CSSM while handing coherent WLFM signals well, a novel method based on instantaneous beamspace array model is developed to estimate the modulation parameters and 2-D AOA of the WLFM signals, and without spatial smoothing or focusing, the proposed method can also process coherent WLFM signals effectively.5. The Jacobi-Anger series representation of sinusoidal frequency-modulated signals is first analyzed, then one new approach based on Jacobi-Anger series representation is derived to estimate the modulation parameters and 2-D AOA of the wideband sinusoidal frequency-modulated signals.6. The problem of array error calibration in nonstationary signal environment is discussed, and in direction-dependent sensor perturbation model, one novel method which can offer closed-form 2-D AOA estimation and model errors self-calibration, is proposed to estimate 2-D AOA of wideband chirp signals, by exploiting the algebraic structure of wideband chirp signal’s time-varying manifold.更多还原