【作者】 钟舜聪；

【导师】 丁康；

【作者基本信息】 汕头大学 ， 机械电子工程， 2002， 硕士

【摘要】 论文是以国家自然科学基金项目“机械振动信号处理中的频谱校正理论”以及国家教育部高等学校骨干教师资助计划项目“机械振动信号处理中一些关键理论的研究”为依托，开展对频谱校正技术的研究以及动态信号分析虚拟仪器软件系统的研制。 针对由于能量泄漏造成的频谱误差，深入研究了密集频谱的频域参数识别法，从理论上加以推导，并通过仿真证明了该方法的有效性，可以识别1个频率分辨率以内的两临近频率分量。综合分析现有三种离散频谱相位差校正方法，推荐在小噪声工程实际中加Hanning窗的第一种相位差法(采连续两段信号作相同点数FFT的相位差校正法)进行校正。在研究时域平移的离散频谱相位差校正方法和改变窗长离散频谱相位差校正方法的基础上，发现这几种离散频谱相位差校正法的基本原理是一致的，就是通过时移和加不同的对称窗进行两次FFT分析，并利用离散频谱对应峰值谱线的相位差以求得频率和相位校正量。利用这一原理提出通用的离散频谱相位差校正方法：时域平移+改变窗长+改变窗函数，即第二段时域序列比第一段滞后L点，采用不同窗函数对这两段时域分别作N点和M点的FFT分析。仿真结果表明，该方法实现方便，校正精度高，适合各种对称窗函数，抗噪声能力强。 研究了基于Windows98的高速数据采集的方法、MATLAB&Visual Basic以及MATLAB&Visual C++混合编程的思想和方法。并采用动态信号离散频谱分析领域最新研究成果，基于虚拟仪器设计思想，研制一套方法新、功能全、实时性好、精度高的基于Windows的通用动态信号分析虚拟仪器软件系统。更多还原

【Abstract】 Spectrum correcting technology in dynamic signal analysis and software system are presented in this thesis. The study is based on the project of National Natural Science Fund of China and the State Ministry of Education of China, which deeply research on spectrum correcting technology in dynamic signal analysis and developing dynamic signal analysis virtual instrument software system.In order to avoid spectrum error due to energy leakage, the thesis research on identification of parameters of two close frequency components in discrete spectrum analysis, and simulation result shows the validity of the method. After analysising three existed methods for phase difference correction on discrete spectrum, it is recommended to correct the spectrum containing small noise in engineer practice by the first phase difference correcting method (performing FFT analysis on two continuous original sequence) with Harming window. Study shows that the basic principles of three existed methods for phase difference correction on discrete spectrum are identical, by which the twice FFT analysis through time-domain shifting time series or changing window’s length is performed, and the spectrum by making use of the phase difference of two corresponding peak lines are finally corrected. Consequently, a synthesized method is presented in this thesis, which includes time-domain shifting ,window-length changing and window-type changing (the second time-domain sequence lags behind the first one with L points, performing N-point FFT analysis on the first sequence and M-point FFT analysis on the second one with different windows, and finally correcting peak lines). The simulation result shows that the synthesized method is easily implemented with high precision, which is also applicable for all kinds of symmetrical window function and having high ability of anti-noise.This thesis also study on the method of high speed data acquisition based on windows98, mixed programming by Visual C++ & Matlab and Visual Basic & Matlab. Adopting the latest research result on dynamic signal discrete spectrum analysis and the idea of virtual instrument, a dynamic signal analysis virtual instrument software system is developed, which is based on windows and has good performance.更多还原