Feature Extraction Method For Weak Modulation Of Ship Radiated Noise

Sound waves are the only form of energy currently known to be effectively transmitted over long distances in the ocean.A sonar is a device that detects and locates hydroacoustic targets over long distances using underwater sound waves.The passive sonar detects,locates and classifies waterborne acoustic targets by listening to the ship’s radiated noise signals underwater and extracting target characteristics from the signals.The broadband modulation feature reflects physical parameters such as the number of propeller blades and spindle speed of the ship’s target propeller.In a sonar system,the Detection of Envelope Modulation on Noise(DEMON)method is used to obtain broadband modulation characteristics of naval radiation noise.At present,the increasing development of vibration and noise reduction technology makes the radiation noise intensity and characteristics of hydroacoustic targets continue to weaken,making it increasingly difficult for conventional DEMON methodologies to extract the modulation characteristics of ship’s radiation noise in the actual marine environment,limiting the target detection and classification identification performance of the sonar underwater.In this paper,the following studies are carried out for the detection and extraction of weakly modulated features in shipboard radiated noise signals.The generation mechanism of the ship’s radiated noise and the statistical law of the signal are analyzed in this paper.Considering the modulation frequency band distribution of the broadband modulation signal and the noise fluctuation caused by the target maneuver,a fine model of the ship’s radiated noise is established,and a signal simulation method is designed.The principle of demodulation spectrum analysis and line spectrum detection of broadband modulation signals by DEMON algorithm is analyzed.A broadband compensation method for frequency band compensation is proposed for the band gain loss caused by the frequency division of the sub-band demodulation algorithm.This method is designed according to the modulation intensity of each frequency band.The frequency band compensation filter weights each frequency band of the original signal,and then performs full-band demodulation to obtain the DEMON spectrum.Experiments show that this method can obtain higher quality DEMON spectrum than sub-band demodulation.The cyclostationary characteristics of the propeller noise are analyzed,and the characterization of the modulation function and modulation frequency band distribution in the broadband modulation signal in the cyclic spectrum is analyzed.For weakly modulated signals,a demodulation method(Cyclostationary Spectrum Entropy Algorithm,CSEA)based on cyclic spectrum and spectral entropy is proposed.Spectral entropy is used to judge the spectral quality of each frequency band of the cyclic spectrum,and a reasonable weighted fusion coefficient is designed to obtain higher signal noise.Than the DEMON spectrum.Experiments show that CSEA’s relative sub-band demodulation algorithm can effectively improve the ability to detect weak signals.A Harmonic Transform Algorithm(HTA)is proposed to extract the axial frequency in the demodulation spectrum.The HTA uses the Harmonic-Transform transform to solve the harmonic spectrum of the target signal,and proves that the harmonic spectrum has the maximum value when it is output on the target axis.Harmonic-Transform concentrates the energy of frequency and harmonics together,and has better suppression effect on harmonic loss and interference line spectrum.In addition to the advantages of HTA,which has a good ability to extract the axis frequency for signals with low signal-to-noise,HTA can also interpolate signals with a short sampling time to improve the frequency resolution of the spectrum.HTA obtains a high-precision harmonic spectrum by setting the frequency interval of the conversion matrix,and thus obtains a more accurate axis frequency.