Study On Feature Extraction Technique Of Underwater Small UUV Radiated Noise

With the development of unmanned and intelligent technologies,various types of unmanned devices have been gradually put into use.The underwater unmanned submersible like UUVs faces various needs such as underwater reconnaissance,searching,mapping,etc.,and has developed various types and models adapted to different scenes.The frequency and threat of invading UUV has also gradually increased.It has become a research hotspot for the detection and identification of underwater small targets.Among them,the feature extraction method based on UUV radiation noise signal processing and the analysis of feature will become the key to determine the success rate of small target detection and recognition.This thesis focuses on the research of underwater UU radiation noise feature extraction.Firstly,the UUV radiation noise mechanism is analyzed,and the basic principles and influencing factors of UUV vibration,propeller rotation and UUV and surrounding water body are analyzed.The analysis results show that UUV has the characteristics of line spectrum and modulation spectrum usually possessed by normal underwater vehicles.However,it is greatly affected by the environment and its own working conditions,and the instability in the features is strong.According to the general underwater target radiation noise signal processing and feature extraction method,two sets of measured samples obtained from a 50 kg class light UUV in different water environments were processed.The spectral characteristics and modulation spectrum were obtained by LOFAR,DEMON and high-order spectrum-3/2 spectrum.At the same time,the signal processing method based on the Mel-Frequency Cepstral Coefficients(MFCC)method of human auditory characteristics is studied.The processing results show that UUV radiation noise has the common characteristics of ship radiation noise under high SNR conditions,and the classical feature extraction method is still feasible for small target radiation noise.Finally,aiming at the characteristics of low target radiated noise intensity,large dynamic change,strong instability and large environmental impact,the feature extraction method based on chaos theory is studied.The chaotic sequence entropy,Lyapunov exponent and correlation dimension are analyzed.Feature extraction is used to extract the characteristics of small target radiation noise,and the chaotic feature extraction ability of small target radiation noise characteristics under different SNR is discussed.The simulation study and experimental dataresult chaotic characteristics are used as potential application methods for UUV radiation noise feature extraction..According to the simulation results of the simulated and measured data samples,the UUV radiation noise characteristics can be extracted by the conventional ship target radiation noise signal processing method and feature extraction technology in the stage of stable signal characteristics and high signal-to-noise ratio.In low SNR environment,chaotic characteristics and chaotic features are potential methods for extracting UUV features.The introduction of chaotic domain features complements and enriches the UUV radiation noise feature extraction method.