| The research on target feature extraction based on ship radiated noise has always been one of the research directions in the field of underwater acoustics.At the same time,it is also its focus and difficulty,especially the emergence of quiet submarines,which puts forward higher requirements for target detection and feature extraction under low signal-to-noise ratio.The power spectrum of ship radiated noise is composed of two parts: continuous spectrum and line spectrum.Therefore,in this paper,from the perspective of signal amplitude and signal phase,the research on the feature extraction method of ship radiated noise continuum modulation is carried out to improve the feature extraction ability of continuum target.In addition,the time fluctuation characteristics of the amplitude and phase of the ship radiated noise line spectrum are also used to carry out the research on the extraction method of the low-frequency line spectrum feature of the ship radiated noise,so as to further improve the effect of the line spectrum feature extraction.This article lays the foundation for the subsequent classification and recognition of artificial intelligence targets.The specific work is as follows:First of all,in the study of amplitude modulation feature extraction,which based on the physical generation mechanism of the continuous spectrum of ship radiated noise,a theoretical model of the continuous spectrum of ship radiated noise was established.The traditional envelope demodulation method was expanded and the influence of stability and signal-to-noise ratio factors are discussed.In addition,the effect of modulation depth and demodulation bandwidth in the broadband noise demodulation process is also studied.Finally,the sea trial experimental data further verifies the ability of traditional DEMON spectrum based on amplitude information to extract the characteristics of ship radiated noise.In the study of phase modulation feature extraction,the theory states that the phase modulation feature can more fully characterize the dynamic characteristics of the signal,the group delay function,which represents the characteristics of phase modulation,has been corrected for the defect of "spike effect",which further strengthens the fineness of its characterization of the spectrum structure;By comparing the influencing factors of phase modulation characteristics and amplitude modulation characteristics,such as stability,signalto-noise ratio,demodulation bandwidth,it is proved that the phase modulation characteristics have higher stability,but the ability to suppress noise is weaker.Finally,the sea trial experimental data further verified the ability of the phase modulation characteristic spectrum to extract the characteristics of ship radiated noise.In the study of line spectrum feature extraction,by establishing the theoretical model of the line spectrum signal and observing and analyzing its time-frequency domain data,it is obtained that the amplitude and phase at the frequency point of the line spectrum have regular fluctuation characteristics.Aiming at this characteristic,the theory derives a line spectrum extraction method based on the fluctuation characteristics of the line spectrum.The effects of the signal-to-noise ratio,the overlap and length ratio of the window function are studied by numerical simulation analysis.Compared with the classical method,it is concluded that the line spectrum extraction based on the amplitude and phase fluctuation characteristics has a stronger ability to suppress noise.In addition,it is obtained that it needs a longer window length and overlap ratio to achieve its excellent performance.Finally,through the processing and analysis of sea trial data,the feasibility and effectiveness of the extraction algorithm based on the fluctuation characteristic line spectrum are verified. |
