复杂海洋环境下基于贝叶斯聚焦法的目标跟踪算法研究

Environmental uncertainty often represents the limiting factor in matched-field localization. Within a Bayesian framework, Bayesian focalization has been widely applied to solve the augmented localization problem, in which the environmental parameters, source ranges and depths are considered to be unknown variables. However, the position of the moving source changes with time, which limits the observation interval and the number of acoustic signals. We have to estimate lots of unknown parameters by limited observation information. When the target is variable motion with low noise or the environment has great uncertainty, Bayesian focalization gets worse..This project will research the target with unknown movement in the complex ocean environment, and analyze the parameter sensitivity and the parameter coupling problem. We will also research the space and time correlation radius, and the target motion form, in order to build the window function. Acoustic signals from a series of observations are treated in a simultaneous inversion. This provides the most informative solution, since data from multiple source locations are brought to bear simultaneously on the environmental unknowns, which in turn better constrain source locations. This project introduces time-unvarying parameters to describe the source position, and the source positions are inverted indirectly by the time-unvarying parameters, which reduces the estimated parameter dimension effectively. Kalman gain is confirmed, and the water depth, sound speed, seabed parameters and source positions are dynamically estimated by recursive form..This project will expand the application of Bayesian focalization method, weaken the influence of environment mismatch and target motion mismatch to source localization, and provide a new and effective mean to track and locate a variable motion target in the complex environment.

海洋环境参数失配导致定位性能大幅度下降是匹配场定位所面临的难题之一，应用贝叶斯理论对环境聚焦，是当前解决该难题的有效途径。然而，运动声源的位置时变性限制了观测时间长度和观测信息量，不得不利用有限的观测信息实现众多参数的估计。海洋环境变化复杂、目标辐射噪声降低且运动规律多变，成为制约该方法推广应用的主要瓶颈。本项目以复杂海洋环境下运动规律未知的目标为研究对象，理论分析参数敏感性及参数耦合情况；研究不同声场时空相关半径和目标运动形式下的窗函数，将多个时刻的接收信号同时反演，引入能够描述声源位置随时间规律性变化的时不变参数，以较少的参数间接反演多个声源位置；确定卡尔曼增益系数，采用递推形式对海深、声速、海底参数和声源位置等状态量进行动态估计。本项目将拓展贝叶斯聚焦法的应用领域，削弱海洋环境和目标运动规律失配对定位的影响，提供一种复杂海洋环境下对变速运动目标跟踪定位的全新且有效手段。

本项目的研究目标是：拓展贝叶斯聚焦法的应用领域，削弱海洋环境和目标运动规律失配对定位的影响，提供一种复杂海洋环境下对变速运动目标跟踪定位的全新且有效手段。课题组围绕研究目标和年度研究计划开展工作，完成了相应的研究内容，取得了创新性较强的理论成果：重点研究了环境参数、阵列参数和声源参数的敏感性，通过对反演参数的敏感性分析，可以将参数分为敏感性参数和不敏感性参数，用来决定参数反演的先后顺序或者是否可采取间接反演；将海洋环境参数与声源位置一同反演的方法提高了匹配场定位算法的环境宽容性和定位精度。研究发现若进一步使用经验公式，间接反演海底吸收系数和海底密度，那么距离定位精度可以得到进一步提高；处理并分析了舰船辐射噪声实验数据，利用连续数据对目标位置进行时间上的连续估计，即序贯估计，研究表明，序贯估计得到的声源距离更加接近于GPS标称距离；研究表明，采用最大后验概率密度解，还是采用一维边缘概率密度最大值，这两种解略有不同，统计结果表明，最大后验概率密度解的精度稍高；研究表明，相比于正交函数表示，学习字典可以利用少数原子（甚至一个原子）更好的解释声速剖面扰动，字典学习可以提高声速剖面的稀疏性，从而提高声速剖面的反演精度。项目期间共发表期刊论文6篇（其中，SCI论文为5篇），会议论文3篇，培养7名硕士生。