Study On Method Of Harmonic Recovery In Alpha And Gaussian Mixed Noise

Harmonic recovery is a very basic but very important problem in the field of signal processing.It has very important theoretical research significance and practical significance.Gauss distribution plays an important role in the traditional signal processing,mainly because the Gauss distribution satisfies the central limit theorem,but the Gauss distribution itself cannot describe the signal with impact characteristics. ? Stable distribution is a very important stochastic process,? stable distribution of signal with impact properties can be well described Non-Gauss and consistent with many natural data.many scholars at home and abroad for the research of signal processing in ? stable noise background.This paper is to solve the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise.The fractional lower order statistics of ? stable distribution is nonlinear,so the harmonic recovery problem in ? stable noise was studied with difficulty.For the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise,we propose a linear covariance function based on the normalized correlation function and proves the existence and universality of linear normalized correlation function.Using the method of linear correlation function of normalized correlation function instead of the original and fractional lower order statistics and linear normalized correlation function.In this paper,the Pisarenko method is used to study the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise.In this paper,the MUSIC method is used to study the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise.In this paper,the ESPRIT method is used to study the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise.the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise derives its satisfying Yule Walker equation(Yule-Walker)and the total least squares(SVD-TLS)method is used to study the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise.Simulation experiments are conducted to verify and compare the performance of various algorithms for the problem of frequency estimation of harmonic signal in ? Stable and Gauss mixed noise.Theory and experiment show that the method proposed in this paper can effectively solve the research problem of this subject.