应用稀疏表示的宽带相干超分辨方位估计方法

Wideband super-resolution bearing estimation method utilizing sparse representation solver after focusing transformation

针对无源声呐中宽带相干处理对邻近弱目标分辨能力仍不足的问题,提出一种相干聚焦后稀疏空间谱求解的宽带超分辨方位估计方法。首先将不同频点数据相干聚焦到参考频点,构成多测量矢量稀疏求解问题,然后采用二阶锥规划内点算法进行稀疏空间谱拟合,实现了宽带弱目标超分辨方位估计。仿真试验结果表明,该方法在弱目标分辨和方位估计精度方面优于宽带非相干和相干信号子空间高分辨处理方法。32元拖曳阵海试数据的处理结果验证了该方法的优越性能,同时说明了该方法在无源声呐探测中应用的可行性。

To aim at the problem that wideband coherent bearing estimation cannot deal with adjacent weak sources well enough in passive sonar, a super-resolution method utilizing sparse representation solver after wideband focusing transformation was presented. The snapshots at different frequency were transformed into one general representation at the reference frequency. The model was reformulated as a problem of sparse representation for multiple measurement vectors. And then it was solved in a second-order cone programming framework by the interior point algorithm to achieve a super-resolution spatial spectrum. The results of simulations indicated that the proposed method had a better performance in the aspects of weak sources resolution and bearing estimation accuracy than wideband incoherent or coherent signal-subspace methods. The good performance was also verified by the sea trial data of a 32-element towed array sonar, and it also demonstrated that the proposed method was feasible to be applied in passive sonar.