Maximum Entropy Spectral estimation based on accelerating genetic algorithm

Abstract

The purpose of this paper was to solve the problems of spectral peak shifting and line splitting existing in Burg's Maximum Entropy Spectral Analysis method (MESA), to enhance the resolution of entropy spectral, and to increase the adaptability of spectral estimation algorithm to signal length, signal noise ratio and initial phase. A method of accelerating Genetic algorithm based maximum Entropy Spectral estimation method (GES) was proposed, where accelerating genetic algorithm was used to optimize the parameters of MESA and the four equivalent conditions of MESA were used as objective function. Three typical simulation cases indicated that the phenomenon of spectral peak shifting and line splitting were absent in the frequency spectral estimated by GES, and the ability to discriminate two closed frequency was improved. Compared with the traditional MESA methods, GES has good performances in signal processing.