【作者】 范庆辉；

【导师】 李兴国；

【作者基本信息】 南京理工大学 ， 信息与通信工程， 2008， 博士

【摘要】 虚拟样机技术是上世纪九十年代随着信息技术、计算机技术的发展而兴起的一项计算机辅助工程技术,它是一种崭新的产品设计研发方法,它使用计算机仿真模型的数字化设计方法代替真实的物理样机设计。毫米波被动探测技术的军事应用潜力,促使毫米波被动探测技术成为各国竞相发展的探测技术。本文将虚拟样机技术引入毫米波被动探测系统研发中,这不仅可以降低毫米波被动探测系统研发成本,还可以有效的减少研发时间。虚拟样机技术必将推动毫米波被动探测系统的发展。本文围绕虚拟毫米波被动探测系统的组成部分,主要对以下方面进行研究。(1)对毫米波辐射计的主要组成部件、噪声和目标信号生成建模,形成虚拟部件,为虚拟毫米波被动探测系统的建立奠定了基础。(2)研究了毫米波辐射计目标视频直流信号的db5小波系数特征,并以此特征为基础,提出了毫米波辐射计目标视频直流信号的小波阈值去噪算法;针对以上小波去噪算法的缺陷,提出了改进方法。使用这些算法为基础开发了毫米波辐射计目标视频直流信号的去噪模块。(3)利用离散余弦变换(DCT)分析了毫米波辐射计目标视频交流信号的DCT系数特征及高斯白噪声的DCT系数特征后,结合SVM、LSSVM、RVM,提出了毫米波辐射计目标视频交流信号的补偿、去噪算法;借鉴小波阈值去噪思想,提出了DCT域毫米波辐射计目标视频交流信号的去噪方法,改善了毫米波辐射计目标视频交流信号的补偿、去噪算法中去噪效果。使用这些算法为基础开发了毫米波辐射计目标视频交流信号的补偿、去噪模块。(4)研究了FSVM算法中隶属度函数选取方法,在综合考虑了样本与各样本集最小包围球之间的关系后研究了一种改进的隶属度函数,并使用基于此隶属度函数的FSVM算法实现了对毫米波目标辐射信号的识别。将三波束阵列探测系统输出的三个信号融合成一个信号,研究了使用分数阶傅里叶变换实现对融合后输出信号的特征提取,再使用FSVM进行分类的目标识别方法。使用这些算法为基础开发了毫米波目标辐射信号的识别模块。(5)使用主要部件组成虚拟毫米波辐射计系统,在此基础上研究了有无低噪声放大器对毫米波辐射计输出信号的影响,验证了各主要部件建模的可行性,为虚拟毫米波辐射计应用奠定了基础;使用虚拟毫米波辐射计、毫米波辐射计目标视频交流信号的补偿、去噪模块和目标识别模块组成虚拟毫米波被动探测系统。更多还原

【Abstract】 Virtual prototyping technology is a computer-aided engineering technology developed in the 1990s with the evolution of information and computer technology. It is a new research method of product design, which uses the digitization design of computer simulation model to replace the design of real physical prototype.As its potential on military application, millimeter-wave (MMW) passive detection has became a worldwide technology competed by all the countries. In this paper, virtual prototyping technology is introduced into MMW passive detecting system, which not only reducing the research cost, but also cutting down the research time. Virtual prototyping technology will certainly promote the development of MMW passive detecting system.Regarding the every components of virtual MMW passive detecting system, this paper conducts the research mainly on the areas as below.(1) Major components of MMW radiometer and noise/target signal are modeling to form virtual components, which established the foundation of virtual MMW passive detecting system.(2) The db5 wavelet coefficient characteristic of video-frequency direct current (DC) signal in MMW radiometer is researched, and wavelet threshold denoising method is taken on video-frequency DC signal. Moreover, the method is improved based on the defects of wavelet denoising above. Module of video-frequency DC signal of MMW radiometer is explored based on these algorithms.(3) The discrete cosine transform (DCT) is utilized to analysis the video-frequency alternating current (AC) signal of MMW radiometer. Based on the DCT coefficient characteristics of target signal and Gaussian white noise, video-frequency AC signal of MMW radiometer compensation and denoising algorithm is proposed with support vector machine (SVM), least square support vector machine (LSSVM) and relative vector machine (RVM). In view of the thought of wavelet threshold denoising, video-frequency AC signal is denoised in DCT domain, which improved the denoising effect of AC MMW radiometer. Compensation and denoising module of video-frequency AC signal of MMW radiometer is explored based on above algorithms.(4) The selection of membership function in fuzzy support vector machine (FSVM) is studied. Considering the relationship of samples with the minimal sphere surrounding by sample set, an improved membership function is researched and signal recognition is realized based on FSVM. After three output signal of three-beam array detecting system are integrated into one signal, fractional Fourier transform (FrFT) is used in feature extraction of integrated output signal and FSVM is utilized in target identification. Target recognition of MMW radiometer module is explored based on the algorithms.(5) Main components composed of the virtual MMW radiometer system, and the influence to output signal is researched whether there is a low-noise amplifier or not. The feasibility of main components modeling is verified, which established the foundation of application of MMW radiometer. At last, virtual MMW passive detecting system is constituted by virtual MMW radiometer, compensation and denoising module of target video-frequency DC signal and target identification module.更多还原