基于海洋观测网的水下目标监测理论与关键技术研究

Underwater target information processing for monitoring is fundamental for marine resource exploitation and marine space security. The sea area of Zhejiang province is important in both civilian and military fields. However, existing approaches, which lack systematical solutions while ignoring the peculiarities of this sea area, are seriously flawed. Considering the characteristics of the ocean observation networks (OON) while comprehensively incorporating signal processing, statistical inference, data fusion approaches, we propose and develop the basic theories and key technologies for underwater target monitoring using the OON, including: (a) modeling of underwater environment and targets based on the OON, (b) theories for accurately tracking of underwater targets with the collaboration of multiple moving platforms, (c) theories for estimation-aided distributed cooperative detection and intelligent identification of underwater targets, (d) theories and approaches for task-driven joint recognition and tracking of underwater targets under highly-coupled conditions, (e) approaches to performance evaluation of joint identification and tracking without knowing ground truth. After theoretical analyses and simulation experiments, the proposed theories and approaches will be evaluated by laboratory and offshore experiments using the experimental system to be successively constructed. Then the theories and approaches will be optimized according to the evaluation results, so that the performance of the OON in continuously monitoring underwater moving targets can be largely improved. The widely-applicable research results incorporate theories in multiple fields and will make breakthroughs in several technologies. The obtained theories and approaches can build a solid theoretical foundation and provide key technologies for monitoring underwater objects and exploiting marine resources in Zhejiang Province and even the whole country. Thus, this research also has important academic significance and wide application perspectives.

结合海洋开发和海洋空间安全的迫切需求，针对浙江海域的特殊性和海洋观测网的特点，考虑到现有的方法缺乏系统性和针对性，本项目综合运用信号处理、统计推断以及数据融合等多领域知识，重点研究基于海洋观测网的水下目标监测基础理论与关键技术，具体包括：基于海洋观测网的水下环境和目标建模、多动平台协作的水下目标高精度跟踪理论、跟踪辅助的水下目标分布式协同探测和智能识别理论、高度耦合条件下任务驱动的水下目标联合识别与跟踪理论、和真值未知情况下目标联合识别与跟踪性能评估理论与方法。在理论研究和仿真试验的基础上，分批逐步建设实验系统，先后开展室内实验和近海试验，并根据实验结果优化理论方案，最终提升整个网络对水下移动目标的动态持续监测能力。本项目具有多学科交叉、多技术突破、多应用拓展等特色，为推动浙江省乃至全国海洋资源开发及水下目标监测奠定坚实的理论基础并提供关键技术的支撑，具有重要的学术意义和广泛的应用前景。

海洋态势感知技术的研究对于海洋开发和海洋安全极为重要。针对浙江海域的海洋环境特殊性及海洋观测网的特点，研究并提出一种具有通用性和可扩展性的水下目标监测理论与关键技术，不仅可对信息处理理论进行有益的扩充，也具有重要的应用价值。本项目基于海洋观测网针对水下目标监测问题进行研究，取得了如下进展：.1）针对基于海洋观测网的水下环境与目标建模问题，构建了多路径水声传播模型、学生t成对马尔科夫目标运动模型、非高斯环境噪声模型，为后续研究提供了理论基础；.2）针对基于海洋观测网的水下目标高精度跟踪问题，利用水下量测结构的特殊性，提出了基于不相关转换的水下目标高精度定位与跟踪新方法，突破了传统的线性最小均方误差估计器的线性结构制约，提高了水下目标跟踪精度；.3）针对海洋观测网的水下目标探测和识别问题，提出了目标检测与面向识别的特征融合网络模型新方法，构建了非线性和不均匀量测扩展目标的跟踪理论与技术，较好解决了目标的精准探测和识别问题；.4）针对海洋观测网多传感器联合识别与跟踪问题，充分利用了估计与决策间的高耦合性，建立了约束联合决策与估计新框架，提出了基于目标运动多样性的联合识别与跟踪理论以及真值未知的联合跟踪与分类性能评估方法，一体化解决了水下多类目标识别与跟踪强耦合的难题，大幅度提升了性能；.5）针对开放式实验系统与试验研究问题，本项目部署了海洋观测网络室内验证平台，并基于浙江大学舟山摘箬山岛海区海底观测网络构建了近海实验平台，通过仿真及外场试验，验证了所提出的方法及传感器节点数据采集等技术的有效性。.本项目取得的进展和突破，为提升海洋观测网对水下目标的动态持续监测能力提供了新的理论框架和技术手段，对于推动浙江省乃至全国海洋资源开发、保障海洋安全具有重要意义。所提出的方法可通用于解决空间监视、自主驾驶、智能家居等实际应用中的目标监测问题。