雷达场景识别系统设计

Scene recognition method based on Web simulation

摘要:

为了能在在海上电子对抗战中及时了解敌军的动向与目的，以长短时记忆循环神经网络和知识图谱结合的方法设计了一套雷达场景识别系统；该系统对雷达的行为特征规律进行了研究和识别，并在雷达行为识别的基础上进行了雷达的场景识别；该系统进行了典型的场景知识构建与集成相似性计算，通过捕获的敌方的雷达数据进行场景识别与仿真； 采用了web仿真技术，搭建了一套三维动态场景，根据场景里面雷达工作剧本进行仿真，产生仿真数据，写入MySQL数据库，然后在场景知识库中进行对比分析，经过实验测试识别出了雷达的行为结果与测试时序内的场景结果；通过场景仿真和识别结果分析，最后能够得出准确率90%以上的识别结果，经实仿真推演满足了实战中的需求。该系统相比其他识别方法，能够以更少的数据和资源就能得出一个较高的识别准确率。

Abstract:

In order to understand the movement and purpose of the enemy in the sea electronic countermeasures, a radar scene recognition system is designed by combining the neural network and knowledge map of memory circulation for a short period of time, the system studies and recognizes the behavioral characteristics of radar, and the radar scene recognition is carried out on the basis of radar behavior recognition, the system makes typical scene knowledge construction and integrated similarity calculation, the scene recognition and simulation by capturing the enemy's radar data, and the web simulation technology is adopted. A set of three-dimensional dynamic scene is set up, according to the radar work script in the scene for simulation, the simulation data is generated, written into MySQL database, and then in the scene knowledge base for comparative analysis, after experimental testing to identify the radar behavior results and test time series scene results; Compared with other identification methods, the system can produce a higher recognition accuracy with less data and resources.