【作者】 杨向锋；

【导师】 张效民；

【作者基本信息】 西北工业大学 ， 检测技术与自动化装置， 2005， 硕士

【摘要】 混沌是20世纪最重要的科学发现之一,被誉为继相对论和量子力学之后的第三次物理革命。混沌理论及其应用是当今世界范围内的一个极具挑战性的研究课题,具有广阔的应用前景。水下目标信号检测和识别对发展现代水下武器具有十分重要的意义,水声信号处理是其研究的主要内容之一。将混沌理论应用于水声信号处理,通过计算水声信号的混沌特征参数,实现对水下目标的检测和识别是当前研究的热点。 本文围绕水中目标识别和分类,着重研究了混沌信号处理方法在舰船辐射噪声信号特征提取和识别中的应用。系统阐述了相空间重构的理论基础,比较了现有的主要相空间重构方法,提出了两种基于高阶统计量的相空间重构新方法;介绍了Lyapunov指数、关联维数、h₂熵及自然测度等混沌特征的定义,并对其特征提取算法分别进行了研究;应用这些特征提取方法对Henon混沌系统和Lorenz混沌系统进行了特征提取仿真分析;计算了两类舰船辐射噪声信号的Lyapunov指数、关联维数、h₂熵、自然测度等混沌特征,研究了这些参数的可分性;采用BP网络分类器,对两类目标的关联维数和自然测度特征进行了分类识别实验,实现了两类目标的分类并且获了较高的识别率。更多还原

【Abstract】 Chaos is one of the most important scientific discoveries in the 20th century and it has been regarded as the third revolution in the physics after the theory of relativity and the quantum mechanics. The chaos theory is one of the most challenging research topics in the world and its application has reached a broader area. Underwater target signal detection and feature extraction have a very important effect on the development of modern underwater weapon. It is the focus of underwater acoustics that chaos theory is used in detecting underwater signals, feature extraction, target recognition and classification by means of calculating chaotic characteristic parameters of underwater acoustic signals.With the underwater target recognition and classification as the main topic of this dissertation, chaotic signal processing methods are applied to the feature extraction and the target recognition of ship-radiated noise. The state space reconstruction theory is discussed systematically. The main methods of state space reconstruction now are analysis compared. Two new methods of state space reconstruction are proposed based on the Higher Order Statistics. The definition of Lyapunov exponent , correlation dimensions , h2 entropy and natural measure are introduced and the estimate methods of these chaotic exponents are studied. The simulation of feature extraction for two chaotic system such as Henon system and Lorenz system has been down. The four chaotic exponents for two different types of ships are calculated and there ability of classification is studied. The classification experiment for two different types of ships is down which base on correlation dimensions and natural measure by using BP classifier. The result of the experiment show that the correlation dimensions and natural measure accomplished the classification and the classification rate is higher.更多还原