Research On Feature Extraction Of Underwater Target And Line Spectrum Tracking

The line spectrum of underwater radiated noise not only has stable frequency,but corresponds to the underwater targets' fixed parameters and essential characteristics.So the feature extraction of the line spectrum is a key to detect underwater target.It's a challenge that how to extract effective features of the line spectrum from the underwater radiated noise.Many countries spend lots of resources and manpower to study it.Our country has been studied the feature extraction of the line spectrum for many years and has gain some achievements,but there are still many problems in the application.To extract features of line spectrum components,the following is the main work of this paper:Firstly, the features of underwater radiated noise in the frequency domain are extracted from the power spectrum of underwater targets.To extract the base frequency of propeller shaft,the paper studies the DEMON spectrum analysis,and discusses the detail of some key steps.In addition,the paper also studies the DEMON spectrum analysis of vector signal.Secondly,based on the energy accumulation property of fractional Fourier transform and the similarity between the radiation noise of accelerated propeller shaft and chirp signal,the paper extracts the features from the radiation noise of accelerated propeller shaft in both of the pressure's and vector's domain by fractional Fourier transform,makes the ternary array which is composed by the frequency's slopes as the new feature and verifies the validity of the feature with experiment and short time Fourier transform.Finally,the line spectrum tracking is studied by the algorithm multi-level decision based on minimum cost function in the paper.At the same times,the paper also finds the improved method to solve the defect of original algorithm and verifies its validity by the simulation and experiment.