Time-Frequency Analysis Of Nonlinear Frequency Modulation Signal And Its Application

In modern signal processing, nonlinear frequency modulation signal (NLFM) is one of the representative signal of non-stationary signals, it is wildly used in radar, sonar, speech, geophysics and biological signal analysis areas. And for these frequency varies with time's signals, both traditional time domain representation and frequency domain representation are inadequate to describe the signal appropriately. Time-frequency analysis is an effective tool to analyze and process non-stationary signals, which maps a one-dimensional signal into a two-dimensional function of time and frequency, shows a good distribution picture in the joint time-frequency domain and makes us know about the change of frequency along with the time clearly. In time-frequency domain, the distribution can show clearly the characteristic of a signal which can't be captured in time domain. Using time-frequency energy distribution we can analyze, process any signal and abstract information in given time or frequency.So, time-frequency analysis has been one of the top interests in signal processing, and many methods have been developed for time-frequency analysis, such as Spectrogram, Wigner-Ville distribution, etc. Nevertheless these methods have various deficiencies, such as cross-term interference, fluctuations, time-frequency resolution limitation etc. These shortcomings cause it difficult to interpret the result and limit its further application. This dissertation focus on the related theories and applications of time-frequency analysis, and the main research works and contributions are as follows:1. In this dissertation, we develop a multiple window time-frequency reassignment method based on a weighted average of orthogonal Hermite window functions. We investigate the optimal weights and optimal windows as a function of signal parameters for the multiple window technique. This algorithm not only improves the time-frequency resolution, maintains a sharp location of reassignment but also restrains cross-term interference and reassigned fluctuations, and reduces the statistical fluctuations caused by noise, reassignment at the same time. Simulation results demonstrate that the method is very effective for both the multi- component signals consisted of two nearly closed LFM signals and NLFM signals.2. We also present a threshold multiple window time-frequency based on the method we have developed, in order to analyze signal in the presence of heavy noise. Then adopt peak detection and iterative method clearly estimates the instantaneous frequency of single component constant amplitude NLFM signals in the presence of heavy noise. The approach is robust against noise and simple in computation. It obtains low average mean squared error(MSE), fast convergence velocity and decreases the noise sensitivity at the same time. Simulation results demonstrate the validity of this method.3. This dissertation further probes the application of the method to signal analysis in the presence of three aspect examples: adaptive AM-FM coherent demodulation of NLFM signals, multi-object radar echo signals resolution, and the blind parameter estimation of frequency hopping signals. We have done some research on their features and compared the advantages and disadvantages when they are exerted in practical use. And on this basis, we make out the superiority of time-frequency analysis by means of experiments.