一种基于独立成分分析的水下宽带信号波达方位估计方法

A Novel ICA-based DOA Estimation Method for Underwater Wideband Source

在水下目标特征信息缺乏、声纳阵列孔径有限的条件下,对水下宽带信号开展高分辨率、高精度的波达方位估计研究,具有明显的研究意义和应用价值。利用水下信源统计独立的特点,采用基于独立成分分析的状态一致性变换算法进行方位估计;考虑到水下加性噪声的高斯分布特性,利用小波降噪的预处理技术,消减混合信号中的高斯噪声成分;针对状态一致性变换在低频段的性能退化,提出一种空间重采样修正方法。使用上述基于空间重采样修正的状态一致性变换算法对仿真及海试数据进行处理,目标方位估计结果表明:该方法具有角度分辨率高、需要阵元数少、对先验知识依赖小、低频段性能稳定的优点,相比常规波束形成、子空间法、传统状态一致性变换3类方法更具性能优越性与实用性。

The estimation of high-resolution and high-accurate direction-of-arrive （DOA） of underwater wideband acoustic sources in the cases of limited array aperture and lack of target feature information is studied. Considering the independent statistical characteristics of underwater bandwidth acoustic sources, a state coherence transform （SCT） method based on independent component analysis （ICA） is utilized to obtain DOA estimation. For the Gaussian distribution characteristics of additive noise, the wavelet denois- ing pre-processing is used to reduce the Gaussian components in the mixed signals. A spatial resampling method is proposed to fix the performance degradation of SCT in low-frequency range. In the computer simulation and sea trial, the proposed algorithm has the advantages of higher resolution, fewer sensors, less dependence on the priori knowledge and better low-frequency performance compared with other tradi- tional algorithms. The signal processing results confirm the feasibility and performance superiority of the proposed algorithm.