基于支持向量机的飞机运动模式识别研究

发布时间：2018-01-14 13:17

本文关键词：基于支持向量机的飞机运动模式识别研究　出处：《电子科技大学》2016年硕士论文　论文类型：学位论文

【摘要】：在现今信息化战争中,利用一切探测感知手段探测出有用信息变得至关重要,信息越充足,对制定战略越有帮助,越容易掌握战争中的主动权,并最终赢得胜利。由于电子干扰技术在现代化战争中也得到了广泛应用,因此如何对抗电子干扰,识别出电子干扰源,做出正确的应对战略具有重要意义。为了应对敌方的电子干扰,需要对雷达信号处理相关技术进行更深入的研究然后探索新的方法来解决问题。无源雷达具有隐蔽性好、生存力强等优点,而目前基于无源雷达的目标探测和定位都是根据多站点来解决的,本文为了探索新的方式来处理目标识别问题,研究单个地基无源雷达接收来自飞行过程中机载雷达发射的直达电磁波信号,提出将时频分析和支持向量机结合的方法来处理该信号以正确识别飞机运动模式。本文的研究内容只是对有源干扰的电子干扰机识别的一部分,作为理论研究,主要针对的是对远距离的、机载雷达为全向天线的飞机运动模式的识别,考虑了匀速直线运动、匀速圆周运动、匀速椭圆跑道运动和匀速8字形轨迹运动这四种运动模式。具体工作如下:(1)以匀速直线运动模式、匀速圆周运动模式、匀速椭圆跑道运动模式和匀速8字形轨迹运动模式这四种运动模式,根据雷达方程和实际飞机飞行速度、电磁波载频、立体空间相对位置、噪声等模拟雷达接收机接收信号。(2)在飞机飞行过程中,由于其相对于无源雷达接收站的径向速度在时刻变化,所产生的多普勒频移也会随时间改变,以此为出发点,采用对非平稳信号具有良好分析效果的时频分析方法:希尔伯特-黄变换(HHT),对接收机接收到的电磁信号进行分析,研究不同运动模式下其多普勒频率变化规律。(3)根据不同运动模式下多普勒频率时频谱特点及总体经验模态分解(EEMD)的分解特性,提出以本征模函数的能量占比作为特征向量,然后采集多组样本数据,通过模式识别算法支持向量机(SVM)学习训练最后识别,判断出飞机的运动模式属于哪一种。本文最终通过单个地基无源雷达接收的来自飞行过程中机载雷达发射的直达电磁波信号,结合时频分析和支持向量机的方法完成了对匀速直线运动、匀速圆周运动、匀速椭圆跑道运动和匀速8字形轨迹运动的正确识别,识别率达86.25%,为电子干扰机的识别提供了新的思路。
[Abstract]:In today's information war, it is very important to detect useful information by all means of detection and perception. The more sufficient the information, the more helpful the strategy is and the easier it is to master the initiative in the war. Because the electronic jamming technology has been widely used in the modern war, how to resist the electronic interference and identify the electronic interference source. It is of great significance to make a correct response strategy. In order to deal with the electronic interference of the enemy. It is necessary to do more in-depth research on radar signal processing technology and explore new methods to solve the problem. Passive radar has the advantages of good concealment and strong survivability. At present, the target detection and localization based on passive radar is based on multi-site. In order to explore a new way to deal with the target recognition problem. A single ground-based passive radar is studied to receive direct electromagnetic wave signals from airborne radar during flight. A method combining time-frequency analysis with support vector machine is proposed to process the signal to correctly identify the aircraft motion pattern. The research content of this paper is only a part of the recognition of the active jamming electronic jamming machine as a theoretical study. The aim of this paper is to recognize the motion pattern of a long distance aircraft with an airborne radar as an omnidirectional antenna. The uniform linear motion and uniform circular motion are considered. The four motion modes of uniform speed elliptical runway motion and uniform velocity 8 zigzag trajectory motion. The specific work is as follows: 1) in the uniform speed straight line motion mode, the uniform speed circular motion pattern. According to the radar equation and the actual flight speed, electromagnetic wave carrier frequency and the relative position of three-dimensional space, the four motion modes of uniform speed elliptical runway and uniform velocity 8 zigzag trajectory are used. Noise and other analog radar receivers receive signal. 2) during the flight of aircraft, the Doppler frequency shift will change with time because of the change of radial velocity relative to the passive radar receiving station at the time. As a starting point, a time-frequency analysis method, Hilbert-Huang transform (HHT), is used to analyze the electromagnetic signals received by the receiver. The variation of Doppler frequency in different motion modes is studied. (3) according to the spectrum characteristics of Doppler frequency and the decomposition characteristics of EEMDs under different motion modes. Taking the energy ratio of intrinsic mode function as the eigenvector, and then collecting multiple sets of sample data, the final recognition is done by pattern recognition algorithm support vector machine (SVM) learning and training. Finally, the direct electromagnetic wave signal from airborne radar in flight is received through a single ground passive radar. Combined with time-frequency analysis and support vector machine, the correct recognition of uniform velocity linear motion, uniform circular motion, uniform speed elliptical runway motion and uniform velocity 8-shape track motion is completed. The recognition rate is up to 86.25%. It provides a new idea for the recognition of electronic jammer.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN974

【参考文献】