Research On The Underwater Moving Source Positioning And Separation Quantization Method Of Complex Noise Source

Noise source positioning and separation quantization technology are not only the main method for testing the noise control capability towards huge mechanical structure underwater and its acoustic feature,but primary approach for vibration and noise reduction as well.They could be effective guidance for noise control,providing solid technical support for the development of the huge structure underwater with high invisibility.In a multi-source structure,if the energy of one source is considerably larger than others,its energy has significant influence on the overall noise level.Therefore,decreasing vibration and noise of the source with the largest contribution can be highly effective.To reduce the radiated noise of huge structure underwater,the primary step is to ensure the position and contribution of the sources,then conducting noise reduction with specific target.Thus,the research on the positioning,detecting,separating and quantizing of the noise source on large mechanical structure has important significance.Based on the development of the source positioning and separation quantization theory as well as their application,this paper,taking the background of huge structure underwater with low speed,conducted both theoretical and experimental research on the positioning of coherent and non-coherent moving source,as well as the separation and quantization of coherent and non-coherent noise source.First of all,this paper conducted research on the moving source positioning of single vector hydrophone and double vector hydrophones.Several positioning methods based on single vector hydrophone were introduced in this paper with their positioning accuracy analyzed and compared.A new method using single vector hydrophone in terms of virtual array for positioning was proposed to overcome the obstacles of single vector hydrophone in coherent sources positioning.The effects of virtual element amount and signal-to-noise ratio on the positioning accuracy were analyzed in this paper.The paper also came up with a source positioning method with double vector hydrophones based on correlation function,analyzed the changing law of its positioning accuracy with the frequency and the distance between elements,and optimized the positioning method combining virtual element method with double vector hydrophones.Moreover,based on the analysis of source identification method,research on the hybrid system with mutual influence among sources was implemented.The paper analyzed theapplicable scope of partial coherence and eigenvalue methods and proposed a source identification method using cross cepstrum.Causality relations between noise sources are used for the identification by analyzing the delayed duration of the signal transmission within the system.Theoretical inference and simulation were used to testify the validation of cepstrum in both coherent and non-coherent source identification.Thirdly,the paper analyzed the feasibility of applying blind source separation method in noise source separation under the background of the hybrid system with mutual influence among sources.It proposed a fast independent component analysis with constraint condition aiming at the uncertainty in both amplitude and sorting combined with the features of hybrid system.Simulation has verified the coherent and non-coherent source separation in both instantaneous mixture system and convolution mixed system.Fourthly,analytic hierarchy process method based on partial coherence spectrum was used in noise source quantization combining the result of source separation.Noise source quantization was broken down into three gradations so that the quantization could be conducted within each gradation.The importance of the noise sources were eventually turned into mathematical quantization results.The paper also proposed an assignment law during the construction of matrix on the base of partial coherence spectrum.Last but not the least,experiments were implemented for the source positioning,separation and quantization underwater.Tests were conducted on the moving source positioning with single vector hydrophone virtual array and double vector hydrophone separately to analyze the positioning accuracy of both methods.Cabin model was used to test the separation and quantization effectiveness of the fast independent component analysis with constraint condition and analytic hierarchy process method based on partial coherence spectrum.The experimental output validated the theoretical analysis,which lay the foundation for future application.