Research On Method Of Parameter Estimation For LFM Signals In Impulsive Noise

Linear frequency modulation signal is a kind of typical non-stationary signal,which is widely used in radar,sonar,seismic exploration and so on.The traditional signal processing,that is,the Fourier transform method,transfers the signal from the time domain to the frequency domain re-processing,which can not take into account the signal in the time domain and frequency domain of local information so it is not suitable for non-stationary signal processing.Thus time-frequency analysis method is proposed including short-time Fourier transform,Wigner-Ville distribution,wavelet transform and other time-frequency analysis method that can precisely obtain various information of non-stationary signal.Conventional signal processing generally assumes that background noise obeys the Gaussian distribution,but in engineering applications,background noise often exhibits strong impulsive properties.The alpha stable distribution can describe impulse noise more accurately,so in this thesis,the study of parameter estimation of linear frequency modulation signal under the background of impulse noise is carried out.1.Compared with Gaussian noise,alpha stable noise possesses the characteristic of strong pulses.Consequently,the maximum likelihood estimation theory and conventional time-frequency analysis are no longer applicable.In this thesis,we use a generalized Cauchy distribution to construct a class of loss function for impulse noise,and analyze its corresponding influence function.Then a robust S transform method is proposed based on the S-transform under the maximum likelihood estimation theory framework.The simulation results show that the robust S transform is more accurate than the Myriad filter and the Meridian filter,and the signal parameter estimation can be realized at lower SNRs and the time-frequency gathering of the signal is improved.2.The synchrosqueezing Chirplet transform method and the fractional lower order synchrosqueezing Chirplet transform are proposed.The frequency of LFM signal is linearly related with time.Based on this fact,the thesis proposes a synchrosqueezing Chirplet transform for LFM signal parameter estimation under the framework of synchrosqueezing using Chirplet transform,and generalizes it to impulse noise environment,that is,fractional lower order synchrosqueezing Chirplet transform.The simulation results show that compared with the conventional time-frequency analysis method,the synchrosqueezing Chirplet transform has no cross term interference,and the time-frequency gathering of LFM signal is enhanced while improving the accuracy of LFM signal parameter estimation.