无需信源数的矢量水听器的改进Capon算法研究

Research on Improved Capon Algorithm for Vector Hydrophones with Unknown Number of Signal Sources

摘要:

针对传统子空间类算法在 DOA（Direction of Arrival）估计存在依赖信源数、算法分辨率有限等问题，提出了一种基于矢量水听器改进 Capon 算法（Improved Capon Estimator，ICE），此算法在未知信源数条件下可实现准确 DOA 估计。该方法利用了矢量水听器同步获取声压与振速数据的特性，构建矢量阵列模型；同时在 Capon 类代价函数框架内引入高阶逆协方差矩阵，通过提升矩阵逆的幂次，抑制信号子空间分量对 DOA 估计的干扰，克服传统子空间算法依赖于先验信源数的问题。本方法基于信号与噪声子空间特性差异，以高阶逆矩阵调控适配水下未知信源场景下的 DOA 估计。经数值仿真验证，该方法在信源数未知条件下，能准确实现高分辨率 DOA 估计，相较于 CBF，MUSIC 等对比算法，对相邻目标的分辨能力更强、具备良好的鲁棒性。

Abstract:

In view of the problems of direction of arrival（DOA）estimation in traditional subspace-based algorithms，such as dependence on the number of sources and limited algorithm resolution，an improved Capon estimator（ICE）based on vector hydrophones is proposed. This algorithm can achieve accurate DOA estimation under the condition of unknown number of sources. It takes advantage of the characteristic that vector hydrophones can simultaneously obtain sound pressure and vibration velocity data，and constructs a vector array model. At the same time，it introduces a higher-order inverse covariance matrix within the framework of Capon-type cost functions to enhance the power of matrix inversion，suppress the interference of signal subspace components on DOA estimation，and overcome the problem that traditional subspace algorithms are dependent on prior knowledge of the number of sources. This method is based on the difference in the characteristics of the signal and noise subspaces， and regulates the high-order inverse matrix to adapt to the DOA estimation in the unknown source scenario. Numerical simulations have verified that this method can accurately achieve high-resolution DOA.