Research On The Method Of Acquiring Underwater Radiated Noise Characteristics Of Ships

The characteristics of underwater radiated noise are one of the main characteristics of ships.For example,the power distribution and envelope characteristics of the continuous spectrum of noise on each frequency component,the frequency,amplitude intensity,stability,time-varying characteristics and spectrum combination characteristics of the line spectrum all contain the inherent information of the ship,and the performance is different on different ships.These inherent characteristics are very important physical parameters for detection(recognition)and anti-detection(acoustic stealth).On the one hand,the gradual quietness of all kinds of military ships and the complexity of radiated noise generation and transmission force ships to continuously improve the methods of acquiring underwater radiated noise and its features,so as to continuously improve the target characteristics database and ensure the identification of friend or foe,military and non-military targets.On the other hand,an accurate grasp of the underwater radiated noise can effectively guide the control of the ship’s target characteristics and stealth design,so as to obtain stealth advantages and improve combat effectiveness.In addition,the international and domestic noise emission standards for merchant ships have been put forward one after another,which makes accurate measurement of underwater radiated noise of ships very important.Therefore,it is of great significance to study the method of acquiring characteristics of underwater radiated noise of ships in both military and civil fields.Past for Marine underwater radiated noise feature representation research mainly based on the theory of mathematical model of the radiated noise prediction and mechanism analysis,simplified model noise simulation and so on,more concentrated on the study of radiated noise feature extraction in wavelet analysis,classic methods,such as LOFAR spectrum and DEMON spectrum analysis,from the point of view of detection,increase infinite array size for gain is not realistic,The feature extraction of weak signal is always a technical problem in the field of underwater acoustic detection and recognition.For radiation noise measurement technology research mainly around the single hydrophone and vector hydrophone,vertical and horizontal array law,etc.,the current measurement method is unable to effectively solve the restrain background noise,weak ability of anti multipath effect,poor,testing distance does not meet the conditions result in far field source level calculation error cannot eliminate,the problem such as directivity can not effectively measure,An accurate measurement method to obtain the target characteristics of sound source is urgently needed.In addition,there is little research on the whole chain including radiated noise characterization,feature extraction and generalized acquisition of feature measurement,and the correlation between radiated noise characterization,feature extraction and feature measurement is seldom analyzed and studied.In this paper,the generation mechanism and inherent physical characteristics of ship underwater radiated noise are described from the physical point of view.On the basis of analyzing the parameters of radiated noise that can be detected and identified,the numerical characterization method of ship underwater radiated noise is described in detail.According to traditional experience model can’t accurate matching of ship and amplitude-frequency characteristic of individual radiation mismatch problem,using the experience model is put forward to revise the ship underwater radiation noise amplitude frequency characteristics,established the numerical representation model of ship underwater radiation noise,close to actual ship underwater radiation noise signal,for the study of the radiated noise feature extraction method provided with the first check.Based on the classical Lofar spectrum and Demon spectrum analysis,principal component analysis(PCA)is used to extract the multi-dimensional radiated noise features extracted by Lofar spectrum and Demon spectrum.PCA is optimized as independent component analysis(ICA)to further improve the identification efficiency of underwater radiated noise features of ships.Based on the method of artificial intelligence identification and machine deep learning,the convolutional neural network and machine learning platform for underwater radiated noise feature extraction of ships were constructed,and the joint training sample set of radiated noise simulation signals combined with real ship test signals was constructed,and the real ship data was used to verify the identification results.Based on the research of the ship radiated noise feature extraction methods show that if further enhance radiated noise feature extraction accuracy,at the same time,the arithmetic of continuous improvement must be accurate measurement to obtain the real value of target characteristics(real property),used for all kinds of feature extraction method and the system to provide "standard feature library" compare the sample and algorithm optimization,In this paper,based on the analysis of various radiation noise measurement method,further puts forward a method based on reverberation pool tested sound power radiated noise measurement method,and puts forward the design method of Marine reverberation pool,and the dimension design,sound insulation design and the flow field calculation and simulation research was conducted on the optimization design,sea reverberation pool design scheme is given,It can be used to accurately obtain and establish the target feature library.The effectiveness of the proposed method is verified by simulation and experiment.