虽然多舰协同作战方式可以提高系统的目标检测、识别、跟踪能力，能够使各舰充分掌握作战海域空中、水面、水下的立体态势。然而，对于指挥决策而言，重要的不是各个传感器当前获取的原始数据本身，而是其中蕴含的全局关系以及未来的发展态势。为了快速准确地描述舰船编队海域环境态势，项目按照知识图谱架构，针对舰船编队多源数据的特点，通过非线性滤波、形态学变换、迁移学习、生成式对抗网络以及神经网络技术，融合多源异构数据信息，构建知识网络模型，解决数据降噪、实体抽取、复杂属性抽取、关系抽取和预测等关键技术问题。在此基础上，通过重新组织各传感器信息之间的关系，以结构化方式在原始数据之上构建一层知识系统，实现按指挥者意图从多侧面综合描述作战海域环境态势的目的。

The improved performance of a naval fleet on target detection, recognition and tracking makes each member ship master the whole environmental status of the sea, including the air status, the sea surface status, as well as the underwarter status. However, for the task of command and decision, the relationship and the trend of the information are more important than the original data itself. This proposal adopts the concept of the knowledge graph to efficiently describe the sea status around a fleet. By using non-linear filtering, morphological transformation, transfer learning, generative adversarial networks, and artificial neural networks, the proposal aims at solving data denoising, entity extraction, complex attribute extraction, relationship extraction and prediction. Eventually, by re-arranging the information of individual sensor, a structural knowledge system is constructed. Based on such structural knowledge, the sea status around a fleet is described from various aspects as needed.