为获取水平柔性长线水听器阵列的实际阵形以提高后续阵列信号处理算法的性能,提出了一种阵形估计的新算法。该算法假设辅助声源信号为宽带球面波,通过对各个频率点处的协方差矩阵进行特征分解,构造了一个基于加权子空间拟合思想的拟合函数。利用高斯-牛顿和交替投影算法实现了阵形的离线估计,并对在实际应用中可能会影响算法性能的四种主要的误差源进行了分析。实验结果表明,相对于辅助声源为窄带平面波的算法,新算法放宽了对辅助声源带宽和距离的要求,具有良好的估计精度和更大的宽容性。
To obtain a true array shape of a horizontal flexible long-line hydrophone array and improve the performance of the subsequent signal processing algorithm,a novel algorithm for array shape estimation is proposed.With the assumption that the signal emitted by auxiliary sources is a wideband spherical wave,a fitting function is first constructed based on the theory of weighted subspace fitting through the eigen-decomposition of the covariance matrix at each frequency point.The off-line estimation of a array shape is attained using Gauss-Newton and alternating-projection algorithm.Four main kinds of error sources are analyzed which can deteriorate the performance of this algorithm in practice.Simulation results show that this new algorithm extend the limit of auxiliary sources' bandwidth and distance,and has better performance and robustness than the algorithm in which plane wave and narrowband source is utilized.