鲁棒的高分辨水声成像机理研究

Inorder to obtain the same resolution image at further distance as possible,an efficient way of an underwater acoustical imaging system is the imaging with the high-resolution DOA estimation at low frequency, but this method is known to be sensitive to the error of target model and array error. In this project, a robust signal model, a two-dimensional spatially distributed source model, is proposed to model the acoustical echo of a real large target. The major advantage of this model is that it can characterize simply the spatial intensity of a real target echo in two dimensional space, whereas most high resolution DOA methods only work efficiently with the point target model, which is effective only at very far distance or for a small target.The new model not only considers the spatial correlation between different scattering cells of the target, but also includes the temporal correlation of diferent cells, so it is more robust than the point target model. Some further researches on this model include the simple and decoupled model, the algorithms to estimate these parameters, and their performances at different conditions. Based on two typical arrays, the rectangular planar array and the circular array, an underwater acoustical imaging method with the two-dimensional spatially distributed model will be built up, and a robust imaging method for sparse targets is also one of focus in this project. A simple experiment will be done in a tank to verify the theoretical results of our model and algorithms, and some neccesarry facilities, such as the receiving array and the compressed model of the real target, will be designed with the help of this funding. This project belongs to a fundamental research for underwater acoustical application, and is also an effective support to the existing acoustical imaging systems. The success of this project will bring a new progress for some underwater areas, like defense and underwater detection.

为了实现一定距离下水下目标的高清晰声成像，就必须采用高分辨成像技术。针对已有的高分辨二维声成像技术存在对模型假定和阵列误差敏感、低信噪比下性能差等缺点，本项目提出一个鲁棒的高分辨水声成像技术--基于二维空间分布源模型的声成像方法。该模型的特点是，既考虑空间上点目标和体目标的回波差异，也考虑回波数据时间上和空间上的相关性，提高了高分辨成像的鲁棒性。同时以此模型为基础，开展模型参数估计算法和性能评估研究，以及鲁棒的自适应波束形成估计技术研究。另外以矩形平面阵和圆阵为基础，研究该模型下二维高分辨图像的构建方法以及稀疏目标的鲁棒成像技术。制作试验用的发射和接收基阵、点目标和体目标模具，在消声水池完成试验的数据采集工作，通过数据分析为理论模型验证和工程实践提供支持。本项目是应用基础研究项目，它的研究进一步拓展了高分辨方法在二维水声成像中的应用，在国防、水下探测等领域具有广泛应用前景

项目建立水下声纳探测目标的二维分布源模型，并在不同阵列结构下提出分布源模型的参数估计方法。首先在MSE准则的基础上，通过对采样协方差矩阵对角加载，提出一种鲁棒的自适应波束形成方法。其次依据二维水下目标散射点的相关性，建立二维非相干和二维相干分布源模型，分别用角功率密度函数和角信号密度函数来表征探测目标的空间分布特征；如果目标散射点具有聚集特性，空间分布用高斯分布函数来建模，反之用均匀分布函数来建模；两种分布函数中中心角表示波达方向，角度扩展表示探测目标在空间的扩展特征；并基于平行线阵、均匀面阵等提出中心角和扩展参数的估计算法；项目系统分析算法的估计性能包括角度扩展、参数耦合，实验条件等因素。最后针对空间分布复杂的分布源，项目利用高斯混合模型建立非对称二维非相干和相干分布源模型，并在L阵的基础上提出参数估计算法。