| As computer technology and digital signal processing(DSP) have been applied in instrumentation and measurement more and more, modem electronic instrumentation is forwarded to general platform and intelligent architecture, in which CPU is kernel. On this platform, DSP algorithms not only improve test resolution, but also take place of some instruments, which can accomplish more functions, such as signal generator, signal analyzer, etc. Therefore, it is important to study DSP algorithms applied in modem instrumentation platform. This dissertation presents valuable algorithms based on the research on high resolution analysis of periodic signal, parameter estimation, wavelet analysis for power measurement and instantaneous signal detection, wavelet application for signal processing in radar. Main contribution of this dissertation can be concluded as follows:A new compensable algorithm for improving the accuracy of periodic signal DFT analysis is presented in details. Firstly by discussing in detail whether the sampling theorem is valid for sinusoidal signal, a principle to avoid the leakage when an infinite sinusoidal signal is blocked into a finite data. Secondly a real-time compensable algorithm is proposed. But currently, the methods for periodic signal analysis adopt data window and interpolation, which have computation burden and limitation, it抯 not proper for real-time signal processing. Thirdly the result show that the new algorithm is capable of reducing 50% of the leakage effect in DFT analysis.The problem of estimating parameters of periodic signal in noise is studied. Firstly based on phase interpolation estimator (PIE) algorithm, a technique for obtaining an estimator that has mean square error of order (N3) is presented, which involves only the Fourier components of the time series at three frequencies. Secondly a improved estimator of frequency, amplitude, and phase is introduced thathas asymptotic variance less than 1.65 times the CRLB. But now, the relevant estimators are provided with asymptotic variance error of order (N?, which is calculated complicatedly. Recently, the least squares estimator are computationally simple, are not sensitive to two or more close spectral lines.Power measurement using the wavelet transform is provided. The advantage of using the wavelet transform data directly is that it provided the distribution of power and energy with respect to the individual frequency bands associated with each level of the wavelet analysis. Compared with FFT method, harrnorious analysis is improved by the wavelet transform. On the other hand, a new signal estimating method-wavelet transform is given. Based on that, the spectrum estimation has better distinguishing rate than that of FFT.Based on wavelet transform, a new method for detecting transient signal has been proposed. By using multiscale wavelet transform, the proposed detector works well under low SNR, which can detect transient signal with unknown waveforms and arrival times. The detection statistic based on the wavelet coefficients has been given and the performance of the detector has been evaluated. Theoretical analysis and computer simulations have shown that the proposed scheme is effective.Based on the general instrumentation platform, an automatic testing system based on VXIbus for radar, a method of recognizing targets is presented. By the energy feature vector method, the target recognition rate of the whole system is improved. On the other hand, wavelet analysis is used to detect the arrival time of scattered wave and denoising. The effect of white noise and time jittered noise on the sample of the received signal are removed. Otherwise, the traditional signal processing methods for radar are based on Fourier analysis. It is invalid to detect the singular signal of radar.The algorithms in our dissertation have been implemented on a personal computer (Inter CPU based), the result prove that our algorithms are correct and efficient.
|
PDF Full Text Request