波導Helix陣三維Fourier變換匹配場定位
發(fā)布時間：2018-10-20 09:11
【摘要】：水聲探測的核心問題之一是目標聲源的被動定位問題。被動定位問題本質是逆問題推斷,通過對傳感器陣接收的含有噪聲的目標輻射聲信號在空間和時間上采樣、分析,結合傳播模型和先驗知識等正向知識,推斷目標所在的空間位置。目標輻射聲信號是在一個上下有界,左右無界的三維波導內傳播,是三維聲場,目標位置具有三維屬性。通常使用的水平線陣或垂直線陣都只是在一維聲場上進行采樣,沒有對三維波導進行充分的空間采樣,因而不具備對目標聲源的三維定位能力。水平陣只能進行方位角估計,垂直線陣只能對距離-深度定位。本文研究具備三維聲場采樣能力的雙螺旋陣(Helix)的性質和三維聯(lián)合定位。在聲源被動定位問題中,目標在每個瞬時時刻只會出現(xiàn)在空間中的某個點上,數(shù)學抽象為,目標在空-時上具有Dirac delta函數(shù)性質。被動定位,應當具備與之相適應的能力,在定位結果上,逼近Dirac delta函數(shù)。實現(xiàn)這一能力的定位方法是匹配濾波,波導中的水下目標定位,其拷貝是發(fā)射信號經傳播模型(模擬信道內聲場的正向傳播)產生的模型場信號,與數(shù)據場信號進行匹配得到模糊度表面,最大函數(shù)值對應的坐標即是估計的聲源位置。這樣一種匹配既可以在陣元域內進行,也可以在空間頻率域即波數(shù)域內進行,因為它們是一對可逆Fourier變換,更有趣的是,人們常常工作在波數(shù)域而不是陣元域。實際處理中,陣列對空間的采樣和傳感器對時間波形的采樣總是有限的。根據Gabor不確實性原理,一個域的有限會帶來另一個域的擴展。所以有限采樣對應的譜域表現(xiàn)就是偏置與泄露,在定位結果的模糊度表面上就顯現(xiàn)出旁瓣。為了使結果盡可能地逼近Dirac delta函數(shù),需要在譜域中施以權重作用,將其能量集中在主瓣內而降低旁瓣響應。最佳的權函數(shù)是信道響應函數(shù)的逆,然而逆問題并不總是存在的且不易計算。一個更廣泛而寬容的權函數(shù)是信道響應的伴隨,對應時域就是時反,對應頻域就是共軛,這也導出匹配濾波的概念。本文從數(shù)學空間角度出發(fā),根據完備正交歸一序列與delta函數(shù)之間的完備性關系,利用三維空間Fourier變換,在波數(shù)域進行匹配,近似delta函數(shù),實現(xiàn)水下聲源定位,即波導中空譜估計聲源定位。本文研究空間Fourier變換,尤其是三維Helix陣的空間Fourier變換的特性。以此為基礎,本文依次研究Helix陣及其褪化陣列的空譜特性。之后在波導環(huán)境中,研究Helix陣及其褪化陣空間Fourier變換匹配定位問題,并對比陣元域匹配場定位。最后,在實驗室波導中,設計并實現(xiàn)了三維Helix陣的聲源定位驗證實驗。
[Abstract]:One of the core problems of underwater acoustic detection is the passive location of target sound source. The passive localization problem is essentially an inverse problem. By sampling and analyzing the acoustic signals of the target with noise received by the sensor array in space and time, combining forward knowledge such as propagation model and prior knowledge, Infer the spatial location of the target. The radiated acoustic signal of the target propagates in a three dimensional waveguide with the upper and lower bounds and the left and right unbounded waveguides. It is a three dimensional sound field and the position of the target has three dimensional properties. Usually, the horizontal or vertical linear arrays are only sampled on one dimensional sound field, and the 3D waveguides are not fully sampled in space, so they do not have the ability to locate the target sound source. Horizontal array can only estimate azimuth, vertical linear array can only locate distance-depth. In this paper, the properties of double helical array (Helix) with three dimensional sound field sampling ability and three dimensional joint positioning are studied. In the passive acoustic source localization problem, the target will only appear at a certain point in space at every instantaneous time. The mathematical abstraction is that the target has the property of Dirac delta function in space-time. Passive positioning should have the ability to adapt to the location results, approximate the Dirac delta function. The localization method to achieve this capability is matched filtering, the underwater target location in the waveguide, whose copy is the model field signal generated by the propagation model of the transmitted signal (the forward propagation of the sound field in the analog channel). The ambiguity surface is obtained by matching the data field signal, and the coordinates corresponding to the maximum function value are the estimated sound source position. Such a matching can be carried out either in the array element domain or in the spatial frequency domain or in the wavenumber domain because they are a pair of invertible Fourier transformations. What is more interesting is that people often work in the wave-number domain rather than in the array element domain. In practical processing, the spatial sampling of array and the sampling of time waveform by sensor are always limited. According to the principle of Gabor uncertainty, the finite of one domain leads to the extension of another domain. Therefore, the spectral domain corresponding to the finite sampling is represented by bias and leakage, and the sidelobe appears on the ambiguity surface of the localization result. In order to approximate the Dirac delta function as much as possible, it is necessary to apply the weight in the spectral domain and concentrate its energy in the main lobe to reduce the sidelobe response. The optimal weight function is the inverse of the channel response function, however, the inverse problem does not always exist and is difficult to calculate. A more extensive and tolerant weight function is the adjoint of channel response. The corresponding time-domain is time-inverse and the corresponding frequency-domain is conjugate, which also leads to the concept of matched filtering. From the point of view of mathematical space, according to the completeness relation between complete orthonormal normalized sequence and delta function, using Fourier transform in three dimensional space, matching in wavenumber domain, approximate delta function, the underwater sound source location is realized. That is, waveguide hollow spectrum estimation of sound source location. In this paper, we study the characteristics of spatial Fourier transform, especially the spatial Fourier transformation of 3D Helix matrix. On this basis, the space-spectrum characteristics of Helix array and its chlorinated array are studied in turn. Then, in the waveguide environment, the matching localization problem of Helix array and its fading array space Fourier transform is studied, and the matching field localization in array element domain is compared. Finally, the sound source location verification experiment of 3D Helix array is designed and implemented in the laboratory waveguide.
【學位授予單位】：浙江大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TB566
本文編號：2282697
[Abstract]:One of the core problems of underwater acoustic detection is the passive location of target sound source. The passive localization problem is essentially an inverse problem. By sampling and analyzing the acoustic signals of the target with noise received by the sensor array in space and time, combining forward knowledge such as propagation model and prior knowledge, Infer the spatial location of the target. The radiated acoustic signal of the target propagates in a three dimensional waveguide with the upper and lower bounds and the left and right unbounded waveguides. It is a three dimensional sound field and the position of the target has three dimensional properties. Usually, the horizontal or vertical linear arrays are only sampled on one dimensional sound field, and the 3D waveguides are not fully sampled in space, so they do not have the ability to locate the target sound source. Horizontal array can only estimate azimuth, vertical linear array can only locate distance-depth. In this paper, the properties of double helical array (Helix) with three dimensional sound field sampling ability and three dimensional joint positioning are studied. In the passive acoustic source localization problem, the target will only appear at a certain point in space at every instantaneous time. The mathematical abstraction is that the target has the property of Dirac delta function in space-time. Passive positioning should have the ability to adapt to the location results, approximate the Dirac delta function. The localization method to achieve this capability is matched filtering, the underwater target location in the waveguide, whose copy is the model field signal generated by the propagation model of the transmitted signal (the forward propagation of the sound field in the analog channel). The ambiguity surface is obtained by matching the data field signal, and the coordinates corresponding to the maximum function value are the estimated sound source position. Such a matching can be carried out either in the array element domain or in the spatial frequency domain or in the wavenumber domain because they are a pair of invertible Fourier transformations. What is more interesting is that people often work in the wave-number domain rather than in the array element domain. In practical processing, the spatial sampling of array and the sampling of time waveform by sensor are always limited. According to the principle of Gabor uncertainty, the finite of one domain leads to the extension of another domain. Therefore, the spectral domain corresponding to the finite sampling is represented by bias and leakage, and the sidelobe appears on the ambiguity surface of the localization result. In order to approximate the Dirac delta function as much as possible, it is necessary to apply the weight in the spectral domain and concentrate its energy in the main lobe to reduce the sidelobe response. The optimal weight function is the inverse of the channel response function, however, the inverse problem does not always exist and is difficult to calculate. A more extensive and tolerant weight function is the adjoint of channel response. The corresponding time-domain is time-inverse and the corresponding frequency-domain is conjugate, which also leads to the concept of matched filtering. From the point of view of mathematical space, according to the completeness relation between complete orthonormal normalized sequence and delta function, using Fourier transform in three dimensional space, matching in wavenumber domain, approximate delta function, the underwater sound source location is realized. That is, waveguide hollow spectrum estimation of sound source location. In this paper, we study the characteristics of spatial Fourier transform, especially the spatial Fourier transformation of 3D Helix matrix. On this basis, the space-spectrum characteristics of Helix array and its chlorinated array are studied in turn. Then, in the waveguide environment, the matching localization problem of Helix array and its fading array space Fourier transform is studied, and the matching field localization in array element domain is compared. Finally, the sound source location verification experiment of 3D Helix array is designed and implemented in the laboratory waveguide.
【學位授予單位】：浙江大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TB566
【參考文獻】
相關期刊論文 前2條
1 劉鳳霞;潘翔;宮先儀;;螺旋線陣匹配場三維定位[J];浙江大學學報(工學版);2013年01期
2 潘超;;運用穩(wěn)相法和模糊函數(shù)設計離焦不敏感光學系統(tǒng)[J];黃石理工學院學報;2010年04期
相關碩士學位論文 前2條
1 項楠;復域DSLA空—時四維匹配場定位[D];浙江大學;2013年
2 舒慧;復域DSLA主動時反目標定位研究[D];浙江大學;2013年
，本文編號：2282697
本文鏈接：http://sikaile.net/guanlilunwen/gongchengguanli/2282697.html
