小波变换和高阶统计量（高阶谱）在振动信号检测中的应用

【作者】 姚雪梅；

【导师】 刘增力；

【作者基本信息】 昆明理工大学 ， 通信与信息系统， 2007， 硕士

【摘要】 信号检测方法是信号处理的一个重要研究方向,以前的信号检测方法多是基于高斯噪声背景的讨论,对非高斯噪声中的信号检测研究较少。但是在雷达,通讯等领域的实际问题中,高斯噪声的假设往往难以满足实际情况。以雷达系统为例,由于各种随机性的脉冲噪声对信道的干扰,总的噪声过程将会偏离高斯分布,而呈现出拖尾分布的特性。这时,基于高斯分布假设的经典检测系统的检测性能将严重恶化。本论文的目的是应用高阶统计量和小波包为工具对噪声中的信号检测问题进行研究。文中首先介绍了信号检测的基本理论,其中包括二元假设检测,统计信号检测系统的设计思想,二元假设检测判决准则;对高阶统计量的定义和性质作了介绍,特别指出了高阶统计量对高斯过程不敏感,这是我们利用它进行信号检测的理论依据;介绍了小波基本理论,阐述了正交小波包的根本思想,它直接来源于多分辨分析和正交小波的构造。然后对含噪声信号分别通过经典信号检测模型进行的仿真,通过高阶统计量检测模型的仿真,通过小波包变换检测模型进行仿真,紧接着对仿真数据和图形进行了分析,最后给出结论。最后本文提出了一种新的信号检测的方法—基于高阶统计量的小波包去噪的信号检测方法,阐述了检测的原理以及影响检测性能的主要参数的选取;最后进行仿真给出了基于高阶统计量的小波包去噪的信号检测性能的分析及其相应结论。更多还原

【Abstract】 Signal detection is one of the most important research branches in the signal process. The former detection methods are mostly based on the Gaussian noise, and seldom based on non-gaussian noise. However, in radar and communication fields, the supposed Gaussian noise can not meet the requirements. Taking radar system as an example, the total noises in the communication channels with various random pulse-noise interference, do not meet gaussian distribution and it trails long. Therefore, the performance of the detection system based on gaussian noise is falling down rapidly.The purpose of this dissertation is to study the problem of detection of random signals in noise with higher-order statistics (HOS) and Wavelet-Packet. Firstly, it introduces the basic theory of signal detection, including the duality supposed detection and multiple supposed detection, the design principle for statistic signal detection system; The definition and properties of HOS are introduced. We especially notice that HOS is insensitive to Gaussian process. This is the theoretical basis for us to use it in signal detection; It introduces the basic theory of the wavelet and the fundamental idea of the orthogonal wavelet-packet, which is coming from the Multi-resolutions analysis and the construction of orthogonal wavelet.Secondly, it stimulants the results of various noise through the classical signal detection system, through the HOS detection system, and through the wavelet-packet decomposition detection system. Then it analyses the dates and the wave forms got from the simulation. Last it gives the conclusion for the performance of the three signal detection system.Lastly, a new detection method of signal is presented-an algorithm of de-noising using wavelet packets based on the HOS. Firstly, it gives the theoretic according of the signal detection. Secondly it gives the affecting factors. Last it gets conclusion by the emulation dates.更多还原